Nothing
inst/shiny-examples/myapp/app.R
In SOMEnv: SOM Algorithm for the Analysis of Multivariate Environmental Data
## App corrected 20210211
requiredPackages <- c("dplyr","plyr","openair", "rlist", "shiny", "kohonen", "shinycssloaders", "shinycustomloader")
ipak <- function(pkg){
new.pkg <- pkg[!(pkg %in% installed.packages()[, "Package"])]
if (length(new.pkg))
install.packages(new.pkg, dependencies = TRUE)
sapply(pkg, require, character.only = TRUE)
}
ipak(requiredPackages)
#---- Upload file fino a 100 MB consentito:
options(shiny.maxRequestSize = 100*1024^2) # per consentire upload files fino a 100 MB
#--------------------------- CSS STYLE -----------------------------------------
CSSHeader<-'font-family: Tahoma, Geneva, sans-serif;
font-size: 22px;
letter-spacing: 2px;
word-spacing: 2px;
color: #000000;
font-weight: 400;
text-decoration: none;
font-style: normal;
font-variant: normal;
text-transform: none'
CSS11<-"[type = 'number'] {font-size:1em; height:25px}
[type = 'text'] {font-size:1em; height:25px}
# [type = 'label'] { font-size:80%}
"
CSS12<-'font-size:1em'
CSS13<-'padding-left: 1px;padding-right: 1px;padding-top: 1px;padding-bottom: 1px;margin-top: -10px'
CSS14<-'height:1em;font-size:1em'
CSS15<-'padding-left: 1px;padding-right: 1px'
CSS16<-'font-size:1em;overflow-x: scroll'
#------------------------------------------------------
CSS21<-"[type = 'number'] {font-size:1em; height:25px}
[type = 'text'] {font-size:1em; height:25px}
# [type = 'label'] { font-size:80%}
"
CSS22<-'padding-left: 1px;padding-right: 1px;padding-top: 5px;padding-bottom: 1px;margin-top: 1px'
CSS23<-'font-size:14px'
CSS24<-'height:11px;font-size:1em'
CSS25<-'padding-left: 1px;padding-right: 1px;padding-top: 1px;'
CSS26<-'font-size:1em;overflow-x: auto;overflow-y: auto'
#------------------------------------------------------
CSS31<-"[type = 'number'] {font-size:1em; height:25px}
[type = 'text'] {font-size:1em; height:25px}
# [type = 'label'] { font-size:80%}
"
CSS32<-'font-size:1em'
CSS33<-'align:left;vertical-align: middle'
CSS331<-'font-size:1em;vertical-align: middle;align: center'
CSS34<-'font-size:11px;text-align: center'
CSS35<-'overflow-y: scroll'
#------------------------------------------------------
CSS41<-'overflow-x: scroll;overflow-y: scroll'
CSS42<-'font-size:1em;overflow-x: scroll;overflow-y: scroll'
#------------------------------------------------------
CSS51<-'padding-left: 1px;padding-right: 1px;padding-top: 1px;padding-bottom: 1px;margin-top: 1px'
##########################################################################################################################################################
#
# PAGE DEFINITION
#
##########################################################################################################################################################
ui <- pageWithSidebar(
headerPanel(div(style=CSSHeader,"SOMEnv Graphical User Interface (v 1.1.2)"),windowTitle = "SOMEnv GUI"),
sidebarPanel(width=3,style='background: white; padding: 1px',
fluidRow(style='height:10vh;background: white;padding-top: 1px;padding-bottom: 1px;padding-left: 1px;padding-right: 1px',
img(src='Logo.jpg', align = "center")#---------------------------------------------------------------------------------------- DA RIVEDERE
),# fluidRow END
fluidRow(style='overflow-y:scroll;height:20vh;background: pink;padding-top: 1px;padding-bottom: 1px;padding-left: 1px;padding-right: 1px',
tags$h4("An R package for mining information from multivariate environmental high frequency data by Self-Organizing Map and k-means clustering algorithms"),
tags$h5("Authors: S. Licen, M. Franzon, T. Rodani, P. Barbieri"),
tags$h5("Licence: GPL-3.0"),
tags$h5("Large datasets up to 100 MB are allowed!")
),# fluidRow END
fluidRow(style='overflow-y:scroll;height:60vh;background: mistyrose;padding-top: 5px;padding-bottom: 1px;padding-left: 1px;padding-right: 1px',
h4("Help"),
br(),
tags$p("
", tags$strong("Tab 1: Load Data")," ", tags$br() ,"
Use Browse button to select the data file. Data are imported in the GUI using the ", tags$a(href=" https://www.rdocumentation.org/packages/openair/versions/2.7-4/topics/import","import"),"function from", tags$a(href="https://cran.r-project.org/web/packages/openair/index.html","openair"), "package.
The data must be in table format in a txt file with date variable (or datetime variable)
in the first column (the header must be 'date') and numeric variables in the following columns.
The user has to write the correct datetime format according to the instructions for", tags$a(href="https://cran.r-project.org/web/packages/openair/index.html", "strptime"),"function used in import function.
Column and decimal separator have to be selected as well. Press Load button to load the data.
When loaded, the first six rows (header) and the last six rows (tail) appear in the lower part of the screen thus the user can check
if the data have been uploaded correctly.
Under the above mentioned tables the number of total uploaded rows as well as the number of deleted rows (containing NA values) are presented.
If the date format is not correct and/or one or more columns contain non numeric values some warnings appear.
When the data are correctly loaded the user can move to the next tab.
"),
tags$p("
", tags$strong("Tab 2: SOM training")," ", tags$br() ,"
The SOM training is based on the use of",tags$a(href="https://www.rdocumentation.org/packages/kohonen/versions/3.0.10/topics/supersom", "som"), "function from", tags$a(href="https://cran.r-project.org/web/packages/kohonen/index.html", "kohonen package"), ". The parameter selection for training the SOM has not default values but some heuristic rules by Vesanto et al. (2000) are widely used, these rules have been embedded in the GUI. Thus, once the data are uploaded, the parameters related to that specific data are calculated. The map dimensions are calculated with three possible specifications (“small”, “medium”, ”large”), and the number of epochs and neighborhood radius vary accordingly. Nevertheless the user can choose custom values for every variable for training the SOM. The neighborhood function can be chosen between “gaussian” and “bubble”. For more details about parameter selection see Clark et al. (2020).
Before training the map the data are by default scaled by variable to have zero mean and variance 1. Scaling is used to ensure all variables have the same importance during training, regardless of absolute magnitude. Moreover an initialization matrix for the SOM units is calculated based on the first two eigenvectors of the data as in Vesanto et al. (2000). Three parameters used in the som function from kohonen package are set by default: topology (“hexagonal”), distance measure (“Euclidean”), learning algorithm (“pbatch”).
When the parameters are chosen pressing the Train SOM button starts the training.
When the calculation is finished in the lower part of the screen on the left the first six rows of the normalized codebook (unit vectors of modeled variables) appear as well as a summary of the training parameters (on the right).
Then the user must save the results (a .RData file) using the button Download SOM output.
For details about the file content see the Appendix below.
If needed, the user can evaluate the quantization error (QE) and topographic error (TE) for the trained map using the Evaluate QE and Evaluate TE buttons respectively.
For more information about these parameters see Clark et al. (2020). Afterwards the user can move to the next tab.
"),
tags$p("
", tags$strong("Tab 3: SOM Map")," ", tags$br() ,"
This tab is provided to visualize the results obtained with the SOM training.
First a “SOM_output” .RData file (obtained from the previous tab) has to be loaded using Browse and Load buttons.
When loaded, a summary of the training parameters used is shown on the left. In the lower part of the screen several plots can be visualized, namely:
Heatmaps: on the left part using Plot button the heatmaps of the SOM map are visualized (i.e. the values distribution on the map of each modeled variable). The filling of the hexagons represents the basic statistics (lower outliers = white, quartiles = gray scale, upper outliers = black) of each modeled variable. For the quartiles and outliers evaluation the", tags$a(href="https://www.rdocumentation.org/packages/graphics/versions/3.6.2/topics/boxplot", "boxplot"), "R function is used with default parameters. The grid (rows x columns) in which plotting the heatmaps can be tailored by the user (see controls above the plot). The download table button allows to download the basic statistics for the modeled variables.
SOM map features:"),
tags$ul(
tags$li("umat: the unified distance matrix (U-matrix) represents the distances between the units. It can be represented as green-white scale (“gs”) or grayscale (“bw”). Darker colors represent lower distances. The choice is related to the fact that the U-matrix can be seen as a representation of valleys (green) and mountains (white – snow). The presence of “mountains” can show the separation of clusters of data. For details see Ultsch et al. (1990);"),
tags$li("hits: this graph represents the distribution of the experimental data on the map, as each unit represents one or more experimental data. It can be visualized either as “grayscale” or “blackfilling”. In the latter the filling of the hexagons is proportional to the number of hits. For further details see also Licen et al. (2020a);"),
tags$li("qerrs: this graph represents the distribution of the quantization errors (QEs) of experimental data on the map, the QE accounts for the accuracy of the match of each experimental vector to the corresponding unit to which it is associated (Best Matching Unit ). It can be visualized either as “grayscale” or “blackfilling”. In the latter the filling of the hexagons is proportional to the quantization error relative values. For further details see also Licen et al. (2019).")
),
tags$p("
The download Hits bs and download Qerrs bs buttons allow to download tables containing the basic statistics for each feature respectively.
From this Tab on all the figures produced in the Tabs can be saved either “as is” using the right click button of the mouse (image format) or using the download pdf button in proximity of the figure. The figure dimensions can be chosen selecting Height and Width (in inches) present nearby the figure or, generally, in the Tab.
"),
tags$p("
", tags$strong("Tab 4: Kmeans clustering")," ", tags$br() ,"
In this Tab the user can further clusterize the data operating a second level clusterization on the codebook using a kmeans clustering algorithm (see Vesanto et al. (1999) and Clark et al. (2020) for details). The calculation are operated on the SOM_output .RData uploaded in the previous tab.
In brief for each number of clusters selected (k = from 2 to max) the kmeans clustering algorithm is run for 100 epochs obtaining a splitting of the units and a corresponding total quantization error for the codebook. The above mentioned algorithm can be run multiple times (iterating times) for each k, and the best of these is selected based on sum of squared errors. For each iteration the function selects a new random seed to set the initialization parameters of the algorithm (see Vesanto et al. (2000)). To obtain reproducible results a seed can be set. In this case the algorithm is evaluated only once (one iteration) for each k. Setting a seed results in fixing the random procedure in order to obtain reproducible. Once selected the max number of clusters and the number of iterating times, press the Run button to start the calculation. The evaluation can take several minutes depending on the number of units and experimental variables. Two progress bars are visualized in the lower right corner of the screen to check the progress of the calculation. At the end of the calculation a Davies-Bouldin index plot appears (see Davies-Bouldin (1979)). The best cluster split is in correspondence to the lower value of the index. Then the user must save the results (a .RData file) using the button Download kmeans output. For details about the file content see Appendix below. The N.clusters control displays automatically the best number of clusters according to Davies-Bouldin index, nevertheless the user can also explore the results for a different number of k. Using the Plot button the SOM map with the cluster split is visualized. The cluster numbers are positioned at the units which are the centroids of the clusters respectively. The user can change the cluster colors clicking on the legend rectangles.
The Cluster profiles graph shows respectively:"),
tags$ul(
tags$li("by cluster: modeled experimental variables profiles of units split by clusters represented by box plots. The boxes show the interquartile range, the thick line shows the median (box whiskers are omitted for the sake of the readability of the figure). The normalized values are used to fit all the variables in the same graph. The user can change some graphical parameters using the controls under the graph;"),
tags$li("by variable: each graph shows an experimental variable, as modeled by the SOM algorithm, split by cluster and depicted by boxplots (using the boxplot R function with default parameters). The values are denormalized to allow checking how the experimental variables were modeled. The grid (rows x columns) in which plotting the heatmaps can be tailored by the user (see controls under the plot);"),
tags$li("Exp by cluster: variables profiles of experimental data split by clusters represented by box plots. The boxes show the interquartile range, the thick line shows the median (box whiskers are omitted for the sake of the readability of the figure). The normalized values are used to fit all the variables in the same graph. The user can change some graphical parameters using the controls under the graph;"),
tags$li("Exp by variable: each graph shows an experimental variable, as present in the experimental data, split by cluster and depicted by boxplots (using the boxplot R function with default parameters). The values are denormalized. The grid (rows x columns) in which plotting the heatmaps can be tailored by the user (see controls under the plot).")
),
tags$p("
See Licen et al. (2018) and Licen et al. (2020b) for details.
"),
tags$p("
As the cluster numbering is random, if the user needs a different numbering of the clusters, the control named Cluster numbers: (upper right of the screen) can be used. The number of clusters has to be expressed by a comma delimited sequence of numbers without spaces. Example: if the default numbering is 1,2,3 and the user needs that cluster 1 becomes 2, 2 -> 3 and 3 ->1, the sequence to be written is 2,3,1. The numbering change affects only the plots, it is not saved in the .RData file. To save the custom cluster numbering see Tab 6.
The user can also load a previously obtained Kmeans_output.RData file (providing to previously load the corresponding SOM_output.RData file in Tab 3!!!) to explore the results. This is done checking the Load file? checkbox and loading the selected file. If the checkbox is unchecked again the user can go back to the latest kmeans evaluation.
"),
tags$p("
", tags$strong("Tab 5: Projection")," ", tags$br() ,"
This tab allows to project into the model data which have not been used to build it. The data to be projected must have the same number of columns (variables) as the training data with date variable (or datetime variable) in the first column (the header must be 'date'). The data file is chosen using the Browse button. Press the Load button to load the data. When loaded, the first six rows (header) and the last six rows (tail) appear in the upper right part of the screen thus the user can check if the data have been uploaded correctly. The number of total uploaded rows as well as the number of deleted rows (containing NA values) are presented. Pressing the Projection button the data are normalized with the same parameters as the training data and projected into the model (referred to the .RData files loaded in Tab 3 and Tab 4). Then the user must save the results (a .RData file) using the button named Download Projection. For details about the file content see Appendix below. Results can be explored visualizing them on the SOM map, for the options description (hits and qerrs) refer to Tab 3. The cluster profiles are built as explained for Tab 4 (options Exp by cluster and Exp by variable) but using the projected data. The visualization parameters of the latter are duplicated from those selected in Tab 4.
The user can also load a previously obtained 'Projection_output' .RData file (providing to previously load the corresponding SOM_output .RData file in Tab 3 and Kmeans_output in Tab 4!!!) to explore the results. This is done checking the Load file? checkbox and loading the selected file. If the checkbox is unchecked again the user can go back to the latest projection.
"),
tags$p("
", tags$strong("Tab 6: Daily profiles")," ", tags$br() ,"
The results obtained in the previous tabs can be explored considering their time sequence. In this tab either the results obtained with the training data (from Tab 3) or the projected data (Tab 5) can be visualized. The visualization parameters of the graphs (e.g. colors, cluster numbering, Y range for the cluster profiles, etc..) are duplicated from those selected in Tab 4. The user can choose the time period to be represented using the controls Starting date and Ending date. The user has to check the control Obs/Day which shows the number of observations for each day as guessed by the function. If the number is not correct, it has to be changed accordingly. Example: if the observations (samples) were recorded every 15 min then the number to be put in the control has to be 4 x 24 hours = 96.
The Daily graph shows the cluster percentage frequency for each day (the X labels dimensions can be changed using the control above the graph). The corresponding table can be downloaded using the table button. The overall tab button downloads a table containing the cluster percentage splitting for the full period. The month tab button downloads a table containing the cluster percentage split by month.
On the upper right of the screen a single observation (sample) can be selected. The BMU number, cluster number and quantization error of the sample are shown. The figure below uses the SOM map and cluster profiles of modeled variables as a legend to show the positioning of the sample respect to the evaluated model: the BMU is labelled by a black dot on the SOM map and the (normalized) experimental values of the sample are shown by red dots onto the corresponding cluster profile. The figure can be useful to visualize possible outliers (see Licen et al. (2018) and Licen et al. (2019)). Two buttons allow to download a table containing the cluster number assignment to the map units (Units cluster assignment) and the to the data (Data cluster assignment, either training or projection as selected in control Dataset). The cluster numbering corresponds to that selected by the user in Tab 4 (Cluster numbers: control).
"),
tags$p("
", tags$strong("References")," ", tags$br() ,""),
tags$ul(
tags$li("Clark et al. (2020): Clark, S., Sisson, S.A., Sharma, A. Tools for enhancing the application of self-organizing maps in water resources research and engineering (2020) Adv. Water Resour. 143, art. no. 103676. DOI: 10.1016/j.advwatres.2020.103676", tags$a(href="https://www.sciencedirect.com/science/article/pii/S030917081931139X", "Link")),
tags$li("Davies-Bouldin (1979): D.L. Davies, D.W. Bouldin, A cluster separation measure, IEEE Trans. Pattern Anal. Mach. Intell. 1 (2) (1979) 224–227, http://dx.doi.org/10.1109/TPAMI. 1979.4766909.", tags$a(href="https://ieeexplore.ieee.org/document/4766909","Link"),""),
tags$li("Licen et al. (2018): Licen, S., Barbieri, G., Fabbris, A., Briguglio, S.C., Pillon, A., Stel, F., Barbieri, P. Odor control map: Self organizing map built from electronic nose signals and integrated by different instrumental and sensorial data to obtain an assessment tool for real environmental scenarios (2018) Sens Actuators, B Chem, 263, pp. 476-485. DOI: 10.1016/j.snb.2018.02.144", tags$a(href="https://www.sciencedirect.com/science/article/pii/S0925400518304246","Link"),""),
tags$li("Licen et al. (2019): Licen, S., Cozzutto, S., Barbieri, G., Crosera, M., Adami, G., Barbieri, P. Characterization of variability of air particulate matter size profiles recorded by optical particle counters near a complex emissive source by use of Self-Organizing Map algorithm (2019) Chemometr. Intelligent Lab. Syst., 190, pp. 48-54. DOI: 10.1016/j.chemolab.2019.05.008", tags$a(href="https://www.sciencedirect.com/science/article/pii/S0169743919301376", "Link"),""),
tags$li("Licen et al. (2020a): Licen, S., Cozzutto, S., Barbieri, P. Assessment and comparison of multi-annual size profiles of particulate matter monitored at an urban-industrial site by an optical particle counter with a chemometric approach (2020) Aerosol Air Qual. Res., 20 (4), pp. 800-809.DOI: 10.4209/aaqr.2019.08.0414", tags$a(href="https://aaqr.org/articles/aaqr-19-08-oa-0414", "Link"),""),
tags$li("Licen et al. (2020b): Licen, S., Di Gilio, A., Palmisani, J., Petraccone, S., de Gennaro, G., Barbieri, P. Pattern recognition and anomaly detection by self-organizing maps in a multi month e-nose survey at an industrial site (2020) Sensors, 20 (7), art. no. 1887. DOI: 10.3390/s20071887.", tags$a(href=" https://www.mdpi.com/1424-8220/20/7/1887","Link"),""),
tags$li("Ultsch et al. (1990): A. Ultsch, H.P. Siemon, Kohonen’s self organizing feature maps for exploratory data analysis, in: Proceedings of International Neural Network Conference (INNC’90), Kluwer academic Publishers, Dordrecht, 1990, pp. 305–308."),
tags$li("Vesanto et al. (1999): Vesanto, J., 1999. SOM-based data visualization methods. Intelligent Data Analysis 3, 111–126. doi:10.1016/S1088-467X(99)00013-X", tags$a(href="https://www.sciencedirect.com/science/article/abs/pii/S1088467X9900013X?via%3Dihub", "Link"),""),
tags$li("Vesanto et al. (2000): J. Vesanto, J. Himberg, E. Alhoniemi, J. Parhankagas, SOM Toolbox for Matlab 5, Report A57, 2000, Available at: www.cis.hut.fi/projects/somtoolbox/package/papers/techrep.pdf", tags$a(href="www.cis.hut.fi/projects/somtoolbox/package/papers/techrep.pdf","Link"),"")
),
tags$p("
", tags$strong("Appendix")," ", tags$br() ,"
In this section the content of the output files is described. The files can be imported in R software using the function 'Load workspace' in the File menu. An object called 'x' is loaded. The content is detailed in the following:
SOM_output.RData "),
tags$ul(
tags$li("$ExpClean: experimental data as imported by the software;"),
tags$li("$dataset: normalized experimental data;"),
tags$li("$ScaleVal: values of column scaling;"),
tags$li("$CenterVal: values of column centering;"),
tags$li("$Epochs: number of training epochs;"),
tags$li("$Neigh: neighborhood type;"),
tags$li("$Dims: map dimensions ($Row and $Col);"),
tags$li("$Coord: coordinates of hexagon centers in the map;"),
tags$li("$codebook: prototype vectors;"),
tags$li("$DeCod: denormalized codebook (variables in original units);"),
tags$li("$Umat: Unified Distance Matrix;"),
tags$li("$Hits: number of hits for each prototype;"),
tags$li("$Bmus: Best Matching Unit (BMU) for each experimental vector;"),
tags$li("$Qerrs: quantization error for each experimental vector."),
),
tags$p("
Kmeans_output.RData:
"),
tags$ul(
tags$li("$centroids: a list containing the centroids for each of the centroids number evaluated (k) ( e.g. $centroids[[5]] is a matrix containing the centroids for five clusters);"),
tags$li("$clusNum: dataframe containing the cluster assignment for prototypes (column 2 for two clusters, etc...);"),
tags$li("$ind: a vector containing the DB-index value for each cluster;"),
tags$li("$err: a vector containing the error for each cluster;"),
tags$li("$BCentr: a list containing the BMU for each centroid ( e.g. $BCentr [[5]] is a vector containing the BMU of centroids for five clusters);"),
tags$li("$ClusExp: dataframe containing the cluster assignment for experimental vectors (column 2 for two clusters, etc...);"),
tags$li("$Seed: the seed number used to start the kmeans algorithm and obtain the corresponding solution."),
),
tags$p("
Projection_output.RData:
"),
tags$ul(
tags$li("$OtherClean: projected data as imported by the software;"),
tags$li("$newdata: normalized projected data normalized with the same parameters as experimental data;"),
tags$li("$HitsPROJ: number of hits for each prototype related to projected data;"),
tags$li("$BmusPROJ: Best Matching Unit (BMU) for each projected vector;"),
tags$li("$QerrsPROJ: quantization error for each projected vector;"),
tags$li("$ClusProj: dataframe containing the cluster assignment for projected vectors (column 2 for two clusters, etc...).")
)
)# fluidRow END
),# END sidebarPanel
mainPanel(width=9,style='background: white',
tabsetPanel(
#---------------------------------------------------------------------------------------------------------------------------------------------------------------------
# FIRST TAB
#---------------------------------------------------------------------------------------------------------------------------------------------------------------------
tabPanel("Load data", fluid = TRUE,
# tags$style(CSS21),
fluidRow(width='80%', style='background: aliceblue;',
# Input: File input and help ----
column(5,tags$div(style= 'font-size:1em',fileInput("Experimental", label="Load experimental data:", accept=c("txt"),buttonLabel=div(style= CSS24,"Browse..."),width="100%")),
actionButton(inputId = "go11",label = "Load")
),# column END
column(3,
tags$div(style= 'font-size:1em',textInput("text11", label="Date format:","%Y-%m-%d %H:%M:%S",width="100%")),
tags$div(style='font-size:1em; font-style:italic; align: left',"Variable date must be in the first column"),
tags$div(style='font-size:1em; font-style:italic; align: left',"See openair import function for date format input")
),# column END
############################### CAMBIATO DA QUI: ############################################################
column(2,
tags$div(style= 'font-size:1em',textInput("text12", label="Separator:","\t",width="100%")),
tags$div(style= 'font-size:1em',textInput("text13", label="Decimal:",".",width="100%"))
############################### A QUI. ######################################################################
)# column END
),# fluidRow END
fluidRow(style='height:60vh;background: white;padding-top: 1px;padding-bottom: 1px;padding-left: 1px;padding-right: 1px',
column(5,verticalLayout(tags$h4("Experimental data header"),
tags$div(style= CSS26,tableOutput("contents11")),
tags$div(style= 'font-weight: bold',br()),
tags$div(style= 'font-weight: bold',textOutput("texto11")),
tags$div(style= 'font-weight: bold',textOutput("texto12")))),
column(5,verticalLayout(tags$h4("Experimental data tail"),
tags$div(style= CSS26,tableOutput("contents12")),
tags$div(style= 'font-weight: bold; color:red',textOutput("texto13"))))
)# fluidRow END
), # tabPanel END
#---------------------------------------------------------------------------------------------------------------------------------------------------------------------
# SECOND TAB
#---------------------------------------------------------------------------------------------------------------------------------------------------------------------
tabPanel("SOM training", fluid = TRUE,
# tags$style(CSS21),
fluidRow(style='height:20vh;background: aliceblue;',
tags$style(type='text/css', ".selectize-input { padding: 1px; min-height: 0;} .selectize-dropdown { line-height: 20px; }"),
# Input: SOM map dimensions ----
uiOutput("UIinput21"),
column(2,tags$div(style= 'font-size:1em',selectInput("Size", "Map size:", c("regular","small","big"), selected = "small",width="100%")),
tags$div(style= 'font-size:1em;font-weight: bold',textOutput("texto21"))),
# Input: SOM training parameters ----
uiOutput("UIinput22"),
# Input: SOM training parameters ----
column(2,tags$div(style= 'font-size:1em;',selectInput("Neigh", "Neighborhood:", c("bubble","gaussian"), selected = "gaussian",width="100%")),
uiOutput("UIinput23")),
column(2,actionButton(inputId = "go21",label = "Train SOM"))
),# fluidRow END
fluidRow(style='height:60vh;background: white;',
############################### CAMBIATO DA QUI: ############################################################
column(5,verticalLayout(h4("Codebook header"),
tags$div(style= CSS26,tableOutput("contents21")),
tags$h4("Map quality parameters"),
splitLayout(cellWidths=c("50%","50%"),actionButton(inputId = "go22",label = "Evaluate QE"),
tags$div(style= 'font-size:1em;',verbatimTextOutput("texto23",placeholder=TRUE))),
splitLayout(cellWidths=c("50%","50%"),actionButton(inputId = "go23",label = "Evaluate TE"),
tags$div(style= 'font-size:1em;',verbatimTextOutput("texto24",placeholder=TRUE))),
div(style='font-size:1em; font-style:italic; color:black',"(The TE calculation can take several minutes)"))),
############################### A QUI #######################################################################
column(4,verticalLayout(h4("Summary"),
tags$div(style= 'font-size:1em;',verbatimTextOutput("texto22",placeholder=TRUE)),
downloadButton("down21","Download SOM output")))
)# fluidRow END
), # tabPanel END
#---------------------------------------------------------------------------------------------------------------------------------------------------------------------
# THIRD TAB
#---------------------------------------------------------------------------------------------------------------------------------------------------------------------
tabPanel("SOM Map", fluid = TRUE,
# tags$style(CSS31),
tags$style(type='text/css', ".butt { padding-left: 2px; padding-right: 2px; padding-top: 1px; padding-bottom: 1px; min-height: 0;font-size:1em}"),
fluidRow(style='background: aliceblue;',
# Input: File input and help ----
column(5,tags$div(style= 'font-size:1em',fileInput("SOMoutput", label="Load SOM output:", accept=c("RData"),buttonLabel=div(style= CSS24,"Browse..."),width="100%")),
actionButton(inputId = "go31",label = "Load")
),# column END
column(5,splitLayout(cellWidths = c("20%","80%"),
tags$h4("Summary:"),
tags$div(style= 'font-size:1em;',verbatimTextOutput("texto31",placeholder=TRUE))))
),# fluidRow END
fluidRow(style='overflow-x:scroll;background:white;',
# Graphs: SOM map and cluster profiles ----
column(7,verticalLayout(splitLayout(cellWidths = c("20%","10%","10%","50%","10%"),
tags$h4("Heatmaps"),
actionButton(inputId = "go32",label = "Plot"),
tags$h4(""),
uiOutput("UIinput31"),
tags$h4("")),
tags$div(withSpinner(plotOutput("plot31"),type=6,color="green")),
splitLayout(cellWidths = c("10%","10%","5%","10%","10%","5%","25%","25%"),
tags$div(style= 'font-size:1em; font-weight: bold',"Height(in):"),
tags$div(style= CSS32,numericInput("h31", label=NULL, 5,width="100%")),
tags$h4(""),
tags$div(style= 'font-size:1em; font-weight: bold',"Width(in):"),
tags$div(style= CSS32,numericInput("w31", label=NULL, 7,width="100%")),
tags$h4(""),
downloadButton("down30","download table",class="butt"),
downloadButton("down31","download pdf",class="butt")
)
)),
column(3,verticalLayout(h4("SOM map features"),
tags$div(withSpinner(plotOutput("plot32"),type=6,color="green"))
)),
column(2,verticalLayout(h5("__________________"),
tags$div(style= 'font-size:1em;',selectInput("menu31", label="Feature", c(" ","umat","hits","qerrs"), selected = " ",width="80%")),
uiOutput("UIinput32"),
tags$div(style= 'font-size:1em; font-weight: bold',numericInput("h32", "Height(in):", 3,width="80%")),
tags$div(style= 'font-size:1em; font-weight: bold',numericInput("w32", "Width(in):", 2,width="80%")),
downloadButton("down32","download pdf",class="butt"),
tags$h5(""),
downloadButton("down33","download Hits bs",class="butt"),
tags$h5(""),
downloadButton("down34","download Qerrs bs",class="butt")
))
)# fluidRow END
), # tabPanel END
#---------------------------------------------------------------------------------------------------------------------------------------------------------------------
# FOURTH TAB
#---------------------------------------------------------------------------------------------------------------------------------------------------------------------
tabPanel("Kmeans clustering", fluid = TRUE,
# tags$style(CSS31),
fluidRow(style='background: aliceblue;',
# Input: File input and help ----
column(3,
tags$div(style= 'font-size:1em; font-weight: bold',numericInput("n41", "Max clusters:", 8,min=2,width="80%")),
tags$div(style= 'font-size:1em; font-weight: bold',numericInput("n42", "Iterating times (x 100 epochs):", 5,min=1,width="80%")) ,
splitLayout(cellWidths = c("30%","70%"),
actionButton(inputId = "go41",label = "Run"),
downloadButton("down41","Download Kmeans output",class="butt"))
),# column END
column(8,verticalLayout(splitLayout(cellWidths = c("15%","55%","10%","20%"),
checkboxInput(inputId = "check41",label = "Load file?",value=FALSE),
conditionalPanel(condition = "input.check41 == 1",
tags$div(style= 'font-size:1em',fileInput("KmeansOutput", label="Load Kmeans output:", accept=c("RData"),buttonLabel=div(style= CSS24,"Browse..."),width="90%"))),
conditionalPanel(condition = "input.check41 == 1",
tags$div(style=CSS33,actionButton(inputId = "go42",label = div(style=CSS331,"Load")))),
uiOutput("UIinput41")), # splitLayout END
tags$div(style='font-size:13px; font-style:italic; color:red',"(Remember to load SOM output in SOM map tab!)"),
uiOutput("UIinput42")
)# verticalLayout END
)# column END
),# fluidRow END
fluidRow(style='width:100%; overscroll-x:auto; background: white;',
# Input: File input and help ----
column(style='background: aliceblue',3,verticalLayout(
splitLayout(cellWidths = c("65%","35%"),
tags$div(style= 'font-size:11px',radioButtons(inputId = "choice41",label = "Set seed?",choices=c("Yes","No"),selected="No",inline=TRUE,width="90%")),
conditionalPanel(condition = "input.choice41 == 'Yes'",
tags$div(style= 'font-size:11px',numericInput("Seed", "", value=7,width="100%")))),
tags$h4("DB-index plot"),plotOutput("plot41"))
),# column END
column(3,verticalLayout(tags$h4("SOM map"),
withSpinner(plotOutput("plot42"),type=5,color="gray")
)),
column(6,verticalLayout(fluidRow(column(5,h4("Cluster profiles")),
column(5,div(style= 'font-size:1em',selectInput("menu41", label=NULL, c("by cluster","by variable","Exp by cluster","Exp by variable"), selected = "by cluster",width="90%")))),
withLoader(plotOutput("plot43"),type="text",loader=list(marquee("Calculating...",
direction="down", behavior = "scroll",width = "90%",style="font-size:15px; font-weight: bold; color:gray")))
))
),# fluidRow END
fluidRow(style='height:5vh;background: white;',
# Input: File input and help ----
column(3,splitLayout(cellWidths = c("15%","25%","15%","25%","5%","15%"),
tags$div(style= 'font-size:1em; font-weight: bold',"H(in):"),
tags$div(style= CSS32,numericInput("h41", label=NULL, 5,width="90%")),
tags$div(style= 'font-size:1em; font-weight: bold',"W(in):"),
tags$div(style= CSS32,numericInput("w41", label=NULL, 7,width="90%")),
tags$h4(""),
downloadButton("down42","pdf",class="butt")
)
),
column(2,splitLayout(cellWidths = c("40%","50%","10%"),
tags$h4(""),
downloadButton("down43","pdf",class="butt"),
tags$h4(""))
),
uiOutput("UIinput43")
)# fluidRow END
), # tabPanel END
#---------------------------------------------------------------------------------------------------------------------------------------------------------------------
# FIFTH TAB
#---------------------------------------------------------------------------------------------------------------------------------------------------------------------
tabPanel("Projection", fluid = TRUE,
# tags$style(CSS31),
fluidRow(style='height:30vh;background: aliceblue;',
# Input: File input and help ----
column(3,div(style= 'font-size:1em',fileInput("Other", label="Load dataset:", accept=c("txt"),buttonLabel=div(style= CSS24,"Browse..."),width="100%")),
fluidRow(column(4,actionButton(inputId = "go51",label = "Load")),
column(8,div(style='font-size:1em; font-style:italic; color:red',"(Remember to load SOM output in SOM map tab!)"))),
tags$div(style= 'font-size:1em;font-weight: bold',br()),
tags$div(style= 'font-size:1em;font-weight: bold',textOutput("texto51")),
tags$div(style= 'font-size:1em;font-weight: bold',textOutput("texto52"))
),# column END
column(2,
tags$div(style= 'font-size:1em',textInput("text51", label="Date format:","%Y-%m-%d %H:%M:%S",width="100%")),
tags$div(style='font-size:1em; font-style:italic; align: left',"Variable date must be in the first column"),
tags$div(style='font-size:1em; font-style:italic; align: left',"See openair import function for date format input")
),# column END
column(1,
############################### CAMBIATO DA QUI: ############################################################
tags$div(style= 'font-size:1em',textInput("text52", label="Separator:",",",width="100%")),
tags$div(style= 'font-size:1em',textInput("text53", label="Decimal:",".",width="100%"))
############################### A QUI. ######################################################################
),# column END
column(style='background: white',3,verticalLayout(tags$div(style= 'font-size:1em; font-weight: bold',"Dataset header"),
tags$div(style= 'font-size:1em;overflow: auto;height:25vh',tableOutput("contents51")))
),# column END
column(style='background: white',3,verticalLayout(tags$div(style= 'font-size:1em; font-weight: bold',"Dataset tail"),
tags$div(style= 'font-size:1em;overflow: auto;height:25vh',tableOutput("contents52")))
),# column END
),# fluidRow END
fluidRow(style='height:60vh;background: white;',
column(3,style='background: aliceblue',verticalLayout(actionButton(inputId = "go52",label = "Projection"),
tags$h4("Summary"),
tags$div(style= 'font-size:1em;',verbatimTextOutput("texto53",placeholder=TRUE)),
downloadButton("down51","Download Projection",class="butt"),
br(),
checkboxInput(inputId = "check52",label = "Load projection?",value=FALSE),
conditionalPanel(condition = "input.check52== 1",
tags$div(style= 'font-size:1em',fileInput("OtherOutput", label="Load Projection file:", accept=c("RData"),buttonLabel=div(style= CSS24,"Browse..."),width="90%"))),
conditionalPanel(condition = "input.check52== 1",
tags$div(style=CSS33,actionButton(inputId = "go53",label = div(style=CSS331,"Load")))),
conditionalPanel(condition = "input.check52== 1",br(),
tags$div(style= 'font-size:1em;',verbatimTextOutput("texto54",placeholder=TRUE)))
)),
column(2,verticalLayout(tags$h4("SOM (projection)"),
tags$div(withSpinner(plotOutput("plot51"),type=6,color="green"))
)),
column(2,verticalLayout(tags$h5("__________________"),
tags$div(style= 'font-size:1em;',selectInput("menu51", label="Feature", c(" ","hits","qerrs"), selected = " ",width="80%")),
uiOutput("UIinput51"),
tags$div(style= 'font-size:1em; font-weight: bold',numericInput("h51", "Height(in):", 3,width="80%")),
tags$div(style= 'font-size:1em; font-weight: bold',numericInput("w51", "Width(in):", 2,width="80%")),
downloadButton("down52","download pdf",class="butt"),
tags$h5(""),
downloadButton("down53","download Hits bs",class="butt"),
tags$h5(""),
downloadButton("down54","download Qerrs bs",class="butt")
)),
column(5,verticalLayout(fluidRow(column(5,tags$h4("Clusters (projection)")),
column(5,tags$div(style= 'font-size:1em',selectInput("menu53", label=NULL, c("Proj by cluster","Proj by variable"), selected = "Proj by cluster",width="90%")))),
withLoader(plotOutput("plot52"),type="text",loader=list(marquee("Calculating...",
direction="down", behavior = "scroll",width = "90%",style="font-size:15px; font-weight: bold; color:gray")))),
downloadButton("down55","download pdf",class="butt"))
)# fluidRow END
), # tabPanel END
#---------------------------------------------------------------------------------------------------------------------------------------------------------------------
# SIXTH TAB
#---------------------------------------------------------------------------------------------------------------------------------------------------------------------
tabPanel("Daily profiles", fluid = TRUE,
tags$style(CSS31), # Non toccare!!! Se viene tolto salta formattazione in tab 3 e altre parti (mah...)
fluidRow(style='height:25vh;background: aliceblue;',
# Input: File input and help ----
column(3,tags$div(style= 'font-size:1em;',selectInput("menu61", label="Dataset", c("training","projection"), selected = "training",width="80%")),
uiOutput("UIinput63")
),# column END
column(3,
uiOutput("UIinput61")
),# column END
column(4,
uiOutput("UIinput62"),
verbatimTextOutput("texto61", placeholder = TRUE)
############################### CAMBIATO DA QUI: ############################################################
),# column END
column(2,
tags$h2(""),
downloadButton("down66","Units cluster assignment",class="butt"),
tags$h4(""),
downloadButton("down67","Data cluster assignment",class="butt")
)# column END
############################### A QUI ############################################################
),# fluidRow END
fluidRow(style='height:60vh;background: white;',
column(6,verticalLayout(splitLayout(cellWidths = c("66%","3%","14%","17%"),
tags$h4("Daily graph"),
tags$h4(""),
tags$div(style= 'font-size:1em; font-weight: bold',"Xlab dim:"),
tags$div(style= CSS32,numericInput("x61", label=NULL, 0.75,width="100%"))),
tags$div(withSpinner(plotOutput("plot61"),type=6,color="green")),
splitLayout(cellWidths = c("7%","12%","7%","12%","9%","11%","3%","19%","20%"),
tags$div(style= 'font-size:1em; font-weight: bold',"H(in):"),
tags$div(style= CSS32,numericInput("h61", label=NULL, 5,width="100%")),
tags$div(style= 'font-size:1em; font-weight: bold',"W(in):"),
tags$div(style= CSS32,numericInput("w61", label=NULL, 7,width="100%")),
downloadButton("down61","pdf",class="butt"),
downloadButton("down62","table",class="butt"),
tags$h4(""),
downloadButton("down63","overall tab",class="butt"),
downloadButton("down64","monthly tab",class="butt")
))
),# column END
column(6,verticalLayout(
tags$h4("Sample profile"),
plotOutput("plot62")),
downloadButton("down65","pdf",class="butt")
)# column END
)# fluidRow END
) # tabPanel END
#-----------------------------------------
) # tabsetPanel END
) # mainPanel END
) # pageWithSidebar END
##########################################################################################################################################################
#
# INPUT & OUTPUT
#
##########################################################################################################################################################
# Define server logic to read selected file ----
server <- function(input, output,session) {
#---------------------------------------------------------------------------------------------------------------------------------------------------------------------
# server: FIRST TAB
#---------------------------------------------------------------------------------------------------------------------------------------------------------------------
#------Da qui richiamo input:
############################### CAMBIATO DA QUI: ############################################################
Experimental <- eventReactive(input$go11,{openair::import(input$Experimental$datapath,date="date",date.format=input$text11,
file.type = "txt", header = 1, sep = input$text12, dec=input$text13)})
############################### A QUI. ######################################################################
#--- Elimination of NA values:
ExpClean<-reactive({na.omit(data.frame(Experimental()))})
#------Da qui output:
output$contents11 <- renderTable({head(data.frame(date=as.character(ExpClean()[,1]),ExpClean()[,-1]))},spacing="xs")
output$contents12 <- renderTable({tail(data.frame(date=as.character(ExpClean()[,1]),ExpClean()[,-1]))},spacing="xs")
output$texto11<-renderText({paste("Number of uploaded rows =",as.character(nrow(ExpClean())))})
output$texto12<-renderText({paste("Number of deleted NA rows =",as.character(nrow(Experimental())-nrow(ExpClean())))})
CheckNum<-reactive({sum(sapply(ExpClean()[,c(2:ncol(ExpClean()))], is.numeric))==length(c(2:ncol(ExpClean())))})
output$texto13<-renderText({if (CheckNum()==FALSE){
paste0("WARNING: one or more variables are non-numeric, check your data!")}
})
#---------------------------------------------------------------------------------------------------------------------------------------------------------------------
# server: SECOND TAB
#---------------------------------------------------------------------------------------------------------------------------------------------------------------------
#--- Centering and scaling of experimental data:
prepare<-reactive({scale(ExpClean()[,-1],center = TRUE,scale=TRUE)})
CenterVal<-reactive({attr(prepare(),"scaled:center")})
ScaleVal<-reactive({attr(prepare(),"scaled:scale")})
dataset<-reactive({prepare()}) # matrice da usare per il calcolo della SOM
#--- Default SOM map dimensions:
DimsD<-reactive({SOMEnv::som_dimR(dataset(),type=input$Size)}) #QUESTO dipende da un input "Size", quindi non serve "Restore Button", basta ricambiare "Size" e si torna ai default!!!!
output$UIinput21 <- renderUI({if(input$go11==0) {
tagList(
column(3,div(style= 'font-size:1em',textInput("txtfake21", "Number of SOM map rows:", value="Waiting for file input...",width="90%")),
div(style= 'font-size:1em',textInput("txtfake22", "Number of SOM map cols:", value="Waiting for file input...",width="90%"))))
} else {
tagList(
column(3,div(style= 'font-size:1em',numericInput("Row", "Number of SOM map rows:", value=DimsD()$Row,min=2,width="90%")),
div(style= 'font-size:1em',numericInput("Col", "Number of SOM map cols:", value=DimsD()$Col,min=2,width="90%"))))}
})
#--- SOM map dimensions evaluated from inputs:
Dims<-reactive({list(Row=input$Row,Col=input$Col,munits=input$Row*input$Col)})
output$texto21<-renderText({paste("Number of units =",as.character(Dims()$munits))})
#--- Parameters evaluated from inputs:
Init<-reactive({SOMEnv::som_initR(dataset(),Dims()$Row,Dims()$Col,Dims()$munits)})
dlen<-reactive({nrow(dataset())})
mpd<-reactive({Dims()$munits/dlen()})
#--- Default parameters evaluated from inputs:
EpochsD<-reactive({max(2,(ceiling(mpd()*10)+ceiling(mpd()*40)))}) # da rough+finetune phase by Vesanto
radMaxD<-reactive({max(1,ceiling(max(c(Dims()$Row,Dims()$Col,na.rm=T))/8))}) # da finetune phase by Vesanto
RadDtext<-reactive({paste(radMaxD(),",",1,sep="")})
output$UIinput22 <- renderUI({tagList(
column(2,div(style= 'font-size:1em',numericInput("Epochs", "Number of epochs:", value=EpochsD(),min=2,width="90%")))
)})
output$UIinput23 <- renderUI({tagList(
div(style= 'font-size:1em',textInput("Rad", "Neigh.radius:", RadDtext(),width="90%"))
)})
#--- Parameters rearranged from inputs:
Rad<-reactive({as.numeric(unlist(strsplit(input$Rad, ',')))})
###------------- SOM MAKE
#------- SOM grid
som_grid <- eventReactive(input$go21,{somgrid(xdim = input$Row, ydim=input$Col, topo="hexagonal", toroidal = FALSE,neighbourhood.fct = input$Neigh)})
#--- SOM training:
som_model <- eventReactive(input$go21,{som(dataset(),
grid=som_grid(),
radius=Rad(),
rlen=input$Epochs,
init=Init(),
dist.fcts="euclidean", #fixed
mode="pbatch",#fixed
keep.data = F)})
#--- Training data projection:
Training<-reactive({map(som_model(), dataset())})
Bmus<-reactive({Training()$unit.classif})
Qerrs<-reactive({Training()$distances})
############################### CAMBIATO DA QUI (20201008): ############################################################
#--- Hits
BmusFreq<-reactive({factor(Bmus(), levels=c(1:Dims()$munits))})
TabHits<-reactive({data.frame(table(BmusFreq()))})
Hits<-reactive({TabHits()$Freq})
############################### A QUI (20201008) #######################################################################
#--- Denormalized codebook
codebook<-reactive({data.frame(som_model()$codes)})
DeCod <- reactive({data.frame(t(apply(codebook(), 1, function(r)r*ScaleVal()+ CenterVal())))})
#--- Prototype x and y coordinates
Coord<-eventReactive(input$go21,{SOMEnv::CodeCoord(input$Row,input$Col)})
#--- Umatrix
Umat<-reactive({SOMEnv::som_umatR(codebook(), input$Row,input$Col)})
#--- Output:
############################### CAMBIATO DA QUI: ############################################################
output$contents21 <- renderTable({head(data.frame(Unit=c(1:nrow(DeCod())),DeCod()))},spacing="xs")
############################### A QUI ######################################################################
output$texto22<-renderPrint({
cat("SOM of size ", Dims()$Row, "x", Dims()$Col, sep = "")
cat("\n Number of samples used for training: ",nrow(ExpClean()),sep = "")
cat("\n Number of prototypes(units): ",nrow(codebook()),sep = "")
cat("\n Topology: hexagonal",sep = "")
cat("\n Neighbourhood function: ",input$Neigh, sep = "")
cat("\n Distance measure used: Euclidean",sep = "")
cat("\n Learning algorithm: pbatch",sep = "")
cat("\n Number of training epochs: ",input$Epochs,sep = "")
})
#--- Save list:
OutputRdata<-reactive({list(ExpClean=ExpClean(),dataset=dataset(),ScaleVal=ScaleVal(),CenterVal=CenterVal(),Epochs=input$Epochs,som_model=som_model(),
Neigh=input$Neigh,Dims=Dims(),Coord=Coord(),codebook=codebook(),DeCod=DeCod(),Umat=Umat(),Hits=Hits(),Bmus=Bmus(),Qerrs=Qerrs())})
output$down21<-downloadHandler(filename=function() {paste0("SOM_output_",format(Sys.time(),format="%Y-%m-%d_%H-%M"),".RData")},
content = function(file) {
SOMoutput<-OutputRdata()
list.save(SOMoutput,file=file)}
)
############################### AGGIUNTO DA QUI: ############################################################
#-- Map quality evaluation
QE<-eventReactive(input$go22,{round(sum(Qerrs())/nrow(ExpClean()),digit=6)})
output$texto23<-renderPrint({cat(QE())})
############################### CAMBIATO DA QUI (20201008): ############################################################
TE<-eventReactive(input$go23,{SOMEnv::SOMtopol(as.matrix(dataset()),as.matrix(codebook()),as.matrix(som_model()$grid$pts))})
output$texto24<-renderPrint({cat(TE())})
############################### A QUI (20201008) #######################################################################
#---------------------------------------------------------------------------------------------------------------------------------------------------------------------
# server: THIRD TAB
#---------------------------------------------------------------------------------------------------------------------------------------------------------------------
SOMoutput <- eventReactive(input$go31,{list.load(input$SOMoutput$datapath)})
output$texto31<-renderPrint({
cat("SOM of size ", SOMoutput()$Dims$Row, "x", SOMoutput()$Dims$Col, sep = "")
cat("\n Number of samples used for training: ",nrow(SOMoutput()$ExpClean),sep = "")
cat("\n Number of variables: ",ncol(SOMoutput()$codebook),sep = "")
cat("\n Neighbourhood function: ",SOMoutput()$Neigh, sep = "")
cat("\n Number of training epochs: ",SOMoutput()$Epochs,sep = "")
})
#-- Heatmaps
MxRow<-eventReactive(input$go31,{if (ncol(SOMoutput()$codebook)<=14) {2} else {3} })
MxCol<-eventReactive(input$go31,{
if(ncol(SOMoutput()$codebook)%%MxRow()==0) {ncol(SOMoutput()$codebook)%/%MxRow()
} else {ncol(SOMoutput()$codebook)%/%MxRow()+1}
})
output$UIinput31 <- renderUI({tagList(splitLayout(style= 'overflow-x: hidden',cellWidths = c("25%","25%","25%","25%"),
div(style= 'font-size:1em; font-weight: bold',"Map x row:"),
div(style= CSS32,numericInput("n31", label=NULL, MxRow(),width="80%")),
div(style= 'font-size:1em; font-weight: bold',"Map x col:"),
div(style= CSS32,numericInput("n32", label=NULL, MxCol(),width="80%"))
)
)})
Table30<-function() {SumTable<-SOMEnv::paramQuant(SOMoutput()$DeCod[,1]); ##################################----- de-normalized!
for (i in c(2:ncol(SOMoutput()$DeCod))) {
P<-SOMEnv::paramQuant(SOMoutput()$DeCod[,i]);SumTable<-cbind(SumTable,P[,2])}
colnames(SumTable)<-c("Statistic",colnames(SOMoutput()$DeCod))
SumTable<-data.frame(SumTable)
return(SumTable)
}
output$down30 <- downloadHandler(filename ="Codebook_basic statistics.txt",
content = function(file) {
write.table(Table30(), file, row.names = F,col.names=T,sep="\t",quote=F)}
)
Plot31<- function() {if (is.null(input$SOMoutput)) {return(NULL)
} else {SOMEnv::HexagonsVar(c(input$n31,input$n32),SOMoutput()$codebook,SOMoutput()$Coord,SOMoutput()$Dims$Row,SOMoutput()$Dims$Col)}}
output$plot31<- renderPlot({if (input$go32[[1]] == 0) return(); Plot31()})
output$down31 <- downloadHandler(filename ="Heatmaps.pdf",
content = function(file) {
pdf(file,height=input$h31,width=input$w31)
Plot31()
dev.off()}
)
#-- SOM map features
output$UIinput32 <- renderUI({if(input$menu31==" ") {
tagList(div(style= 'font-size:1em;',selectInput("menu32", label="Plot type", c(" "), selected = " ",width="80%")))
} else if (input$menu31=="umat"){tagList(div(style= 'font-size:1em;',selectInput("menu32", label="Plot type", c("bw","gs"), selected = "gs",width="80%")))
} else if (input$menu31=="hits" | input$menu31=="qerrs"){tagList(div(style= 'font-size:1em;',
selectInput("menu32", label="Plot type", c("grayscale","black filling"), selected = "grayscale",width="80%")))}
})
Switch3<-reactive({paste0(input$menu31,input$menu32)})
Plot32<- function() {switch(Switch3(),
" "={return(NULL)},
"umatgs"={SOMEnv::UmatGraph(SOMoutput()$Umat,SOMoutput()$Dims$Row,SOMoutput()$Dims$Col,colorscale="gs")},
"umatbw"={SOMEnv::UmatGraph(SOMoutput()$Umat,SOMoutput()$Dims$Row,SOMoutput()$Dims$Col,colorscale="bw")},
"hitsgrayscale"={SOMEnv::HexaHitsQuant(SOMoutput()$Hits,SOMoutput()$Coord,SOMoutput()$Dims$Row,SOMoutput()$Dims$Col)},
"hitsblack filling"={SOMEnv::HexaHits(SOMoutput()$Hits,SOMoutput()$Coord,SOMoutput()$Dims$Row,SOMoutput()$Dims$Col)},
"qerrsgrayscale"={SOMEnv::HexaQerrsQuant(SOMoutput()$Bmus,SOMoutput()$Qerrs,SOMoutput()$Coord,SOMoutput()$Dims$Row,SOMoutput()$Dims$Col)},
"qerrsblack filling"={SOMEnv::HexaQerrs(SOMoutput()$Bmus,SOMoutput()$Qerrs,SOMoutput()$Coord,SOMoutput()$Dims$Row,SOMoutput()$Dims$Col)})
}
output$plot32<- renderPlot({Plot32()})
output$down32 <- downloadHandler(filename =function() {paste0("SOM_",input$menu31,"_",input$menu32,".pdf")},
content = function(file) {
pdf(file,height=input$h32,width=input$w32)
Plot32()
dev.off()}
)
#-- Basic statistics
Table31<-function() {SOMEnv::paramQuant(SOMoutput()$Hits)}
output$down33 <- downloadHandler(filename ="Hits_basic statistics.txt",
content = function(file) {
write.table(Table31(), file, row.names = F,col.names=T,sep="\t",quote=F)}
)
Table32<-function() {SOMEnv::paramQuant(SOMoutput()$Qerrs)}
output$down34 <- downloadHandler(filename ="Qerrs_basic statistics.txt",
content = function(file) {
write.table(Table32(), file, row.names = F,col.names=T,sep="\t",quote=F)}
)
#---------------------------------------------------------------------------------------------------------------------------------------------------------------------
# server: FOURTH TAB
#---------------------------------------------------------------------------------------------------------------------------------------------------------------------
# su dati caricati in Third Tab!!!!
Switch41<-function() {switch(input$choice41,
"Yes"={paste0("Yes")},
"No"={paste0("No")})
}
Kmeans <- eventReactive(input$go41,{if (Switch41()=="No"){SOMEnv::kmeans_clustersRProg(SOMoutput()$codebook,k=input$n41,times=input$n42)
} else {SOMEnv::kmeans_clustersRProg(SOMoutput()$codebook,k=input$n41,times=input$n42,seed=input$Seed) }
})
#-- Risultati aggiuntivi:
############################### CORRETTO DA QUI: 20210211 ###############################################
BCentr<-reactive({BmusCentr(Kmeans()$centroids,SOMoutput()$som_model,length(Kmeans()$centroids))})
ClusExp<-function() {D<-rep(0,nrow(SOMoutput()$ExpClean))
for (i in c(2:length(Kmeans()$centroids))) {d<-BmusClus(SOMoutput()$Bmus,Kmeans()$clusNum[,i]);
D<-cbind(D,d)}; return(data.frame(D))}
############################### A QUI. ######################################################################
#-- Output:
OutputKdata<-reactive({list(centroids=Kmeans()$centroids,clusNum=data.frame(Kmeans()$clusNum),ind=Kmeans()$ind,
err=Kmeans()$err,seed=Kmeans()$seed,BCentr=BCentr(),ClusExp=ClusExp())})
output$down41<-downloadHandler(filename=function() {paste0("Kmeans_output_",format(Sys.time(),format="%Y-%m-%d_%H-%M"),".RData")},
content = function(file) {
KmeansOutput<-OutputKdata()
list.save(KmeansOutput,file=file)}
)
#-- If KmeansOutput file is loaded:
Kmeansoutput <- eventReactive(input$go42,{list.load(input$KmeansOutput$datapath)})
KmeansForPlot<-reactive({if(input$check41==FALSE){OutputKdata()} else {Kmeansoutput()} })
#-- Aggiornamento input in base a cluster migliore:
BestClus<-reactive({if(is.null(KmeansForPlot())) {5} else {which(KmeansForPlot()$ind==min(KmeansForPlot()$ind,na.rm=T))+1}})
output$UIinput42 <- renderUI({tagList(
fluidRow(
column(2,div(style= 'font-size:1em',numericInput("n43", label="N.clusters:", BestClus(),min=2,width="80%"))),
column(1,colourpicker::colourInput("col40", label=div(style= CSS34,"Cl 1"), "red",showColour="background",returnName=T,palette="limited")),
column(1,colourpicker::colourInput("col41", label=div(style= CSS34,"Cl 2"), "yellow",showColour="background",returnName=T,palette="limited")),
conditionalPanel("input.n43>=3",column(1,colourpicker::colourInput("col42", label=div(style= CSS34,"Cl 3"), "green",showColour="background",returnName=T,palette="limited"))),
conditionalPanel("input.n43>=4",column(1,colourpicker::colourInput("col43", label=div(style= CSS34,"Cl 4"), "cyan",showColour="background",returnName=T,palette="limited"))),
conditionalPanel("input.n43>=5",column(1,colourpicker::colourInput("col44", label=div(style= CSS34,"Cl 5"), "darkorange1",showColour="background",returnName=T,palette="limited"))),
conditionalPanel("input.n43>=6",column(1,colourpicker::colourInput("col45", label=div(style= CSS34,"Cl 6"), "magenta",showColour="background",returnName=T,palette="limited"))),
conditionalPanel("input.n43>=7",column(1,colourpicker::colourInput("col46", label=div(style= CSS34,"Cl 7"), "aquamarine",showColour="background",returnName=T,palette="limited"))),
conditionalPanel("input.n43>=8",column(1,colourpicker::colourInput("col47", label=div(style= CSS34,"Cl 8"), "sandybrown",showColour="background",returnName=T,palette="limited"))),
column(2,div(style=CSS33,actionButton(inputId = "go43",label = div(style=CSS331,"Plot"))))
) # fluidRow END
)})
TextClus<-reactive({if(is.null(KmeansForPlot())) {paste(c(1:5),collapse=",",sep="")} else {paste(c(1:input$n43),collapse=",",sep="")}})
output$UIinput41 <- renderUI({tagList(
div(style= CSS32,textInput(inputId = "text41",label = "Cluster numbers:",value=TextClus(),width="90%"))
)})
#-- Output DB-index plot:
Plot41<-function(){if (is.null(KmeansForPlot())) {return(NULL)
} else {
opar <- par(mar=c(4,3,1,1),oma=c(1,1,1,1))
plot(c(2:(length(KmeansForPlot()$ind)+1)),KmeansForPlot()$ind,col="gray",xlim=c(1,length(KmeansForPlot()$ind)+2),type="h",
ylim=c(min(KmeansForPlot()$ind,na.rm=T)-0.05*min(KmeansForPlot()$ind,na.rm=T),max(KmeansForPlot()$ind,na.rm=T)+0.05*max(KmeansForPlot()$ind,na.rm=T)),
xlab="Cluster number",ylab="DB-index",xaxt="n")
text(c(2:(length(KmeansForPlot()$ind)+1)),KmeansForPlot()$ind,round(KmeansForPlot()$ind,digit=2),cex=0.75,pos=3)
axis(1,at=seq(2,length(KmeansForPlot()$ind)+1,1),labels=NA,tcl=-0.3,cex.axis=0.6)
axis(1,at=seq(2,length(KmeansForPlot()$ind)+1,1),labels=seq(2,length(KmeansForPlot()$ind)+1,1),lwd=0,line=-0.2,cex.axis=1)}
on.exit(par(opar))
}
output$plot41<-renderPlot({if (is.null(KmeansForPlot())) {return(NULL)
} else {Plot41()}})
output$down42 <- downloadHandler(filename =function() {paste0("DB-index_plot",".pdf")},
content = function(file) {
pdf(file,height=input$h41,width=input$w41)
Plot41()
dev.off()}
)
#-- Output SOM cluster map plot:
colSeq4<-reactive({rep_len(c(input$col40,input$col41,input$col42,input$col43,input$col44,input$col45,
input$col46,input$col47), input$n43)})
numSeq41<-reactive({as.numeric(unlist(strsplit(input$text41, ',')))})
numSeq42<-reactive({as.numeric(unlist(strsplit(input$text42, ',')))})
numSeq43<-reactive({c(input$y41,input$y42)})
Centroids4<-reactive({data.frame(KmeansForPlot()$centroids[[input$n43]][order(numSeq41()),])})
############################### CORRETTO DA QUI: 20210211 ###############################################
clusNum4<-reactive({SOMEnv::NClusChange(KmeansForPlot()$clusNum[,input$n43],Centroids4(),numSeq41())})
BCentr4<-reactive({KmeansForPlot()$BCentr[[input$n43]][order(numSeq41())]})
ClusExp4<-reactive({SOMEnv::NClusChange(KmeansForPlot()$ClusExp[,input$n43],Centroids4(),numSeq41())})
############################### A QUI. ######################################################################
Plot42<- function() {if (is.null(KmeansForPlot())) {return(NULL)
} else if (is.null(SOMoutput())) {return(NULL)
} else if (length(numSeq41())!=input$n43) {return(NULL)
} else {
opar <- par(mar=c(1,1,1,2))
SOMEnv::HexagonsClus(Centroids4(), clusNum4(), BCentr4(),
SOMoutput()$Coord, SOMoutput()$Dims$Row, SOMoutput()$Dims$Col, colSeq = colSeq4())}
on.exit(par(opar))
}
output$plot42<-renderPlot({if (is.null(KmeansForPlot())|input$go43==0 ) {return(NULL)
} else if (is.null(SOMoutput())) {return(NULL)
} else {Plot42()}
})
output$down43 <- downloadHandler(filename =function() {paste0("SOM map",".pdf")},
content = function(file) {
pdf(file,height=input$h41,width=input$w41)
Plot42()
dev.off()}
)
#-- Output cluster profiles plot:
output$UIinput43 <- renderUI({if(input$menu41=="by cluster"| input$menu41=="Exp by cluster") {
tagList(
column(7,splitLayout(cellWidths = c("8%","7%","10%","7%","10%","7%","12%","10%","12%","20%"),
h5(""),
div(style= 'font-size:1em; font-weight: bold',"Y min:"),
############################### CORRETTO DA QUI: 20210211 ###############################################
div(style= CSS32,numericInput("y41", label=NULL, -2,width="100%")),
div(style= 'font-size:1em; font-weight: bold',"Y max:"),
div(style= CSS32,numericInput("y42", label=NULL, 2,width="100%")),
############################### A QUI. ######################################################################
div(style= 'font-size:1em; font-weight: bold',"Y grid:"),
div(style= CSS32,textInput(inputId = "text42",label=NULL,value="-1,0,1",width="100%")),
div(style= 'font-size:1em; font-weight: bold',"Xlab dim:"),
div(style= CSS32,numericInput("x41", label=NULL, 0.75,width="100%")),
downloadButton("down44","pdf",class="butt")))
)
} else if(input$menu41=="by variable"| input$menu41=="Exp by variable"){
tagList(
column(7,splitLayout(cellWidths = c("15%","15%","15%","15%","15%","25%"),
h5(""),
div(style= 'font-size:1em; font-weight: bold',"Map x row:"),
div(style= CSS32,numericInput("n44", label=NULL, MxRow(),width="80%")),
div(style= 'font-size:1em; font-weight: bold',"Map x col:"),
div(style= CSS32,numericInput("n45", label=NULL, MxCol(),width="80%")),
downloadButton("down44","pdf",class="butt")))
)}
})
############################### CORRETTO DA QUI: 20210211 ###############################################
Plot43<- function() {if (is.null(KmeansForPlot())) {return(NULL)
} else if (is.null(SOMoutput())) {return(NULL)
} else if (length(numSeq41())!=input$n43) {return(NULL)
} else {switch(input$menu41,
"by cluster"={SOMEnv::BoxUnits(SOMoutput()$codebook,clusNum4(),Centroids4(),Ylim=numSeq43(),pitch=numSeq42(),xdim=input$x41)},
"by variable"={SOMEnv::BoxClus(c(input$n44,input$n45),SOMoutput()$DeCod,clusNum4(),Centroids4())},
"Exp by cluster"={SOMEnv::BoxUnits(SOMoutput()$dataset,ClusExp4(),Centroids4(),Ylim=numSeq43(),pitch=numSeq42(),xdim=input$x41)},
"Exp by variable"={SOMEnv::BoxClus(c(input$n44,input$n45),SOMoutput()$ExpClean[,-1],ClusExp4(),Centroids4())})}
}
############################### A QUI. ######################################################################
output$plot43<-renderPlot({if (is.null(KmeansForPlot())|input$go43==0 ) {return(NULL)
} else if (is.null(SOMoutput())) {return(NULL)
} else {Plot43()}
})
output$down44 <- downloadHandler(filename =function() {paste0("Cluster profiles",".pdf")},
content = function(file) {
pdf(file,height=input$h41,width=input$w41)
Plot43()
dev.off()}
)
#---------------------------------------------------------------------------------------------------------------------------------------------------------------------
# server: FIFTH TAB
#---------------------------------------------------------------------------------------------------------------------------------------------------------------------
#------Da qui richiamo input Other:
############################### CAMBIATO DA QUI: ############################################################
Other <- eventReactive(input$go51,{openair::import(input$Other$datapath,date="date",date.format=input$text51,
file.type = "txt",header = 1,sep = input$text52,dec=input$text53)})
############################### A QUI #######################################################################
#--- Elimination of NA values:
OtherClean<-reactive({na.omit(data.frame(Other()))})
#------Da qui output:
output$contents51 <- renderTable({head(data.frame(date=as.character(OtherClean()[,1]),OtherClean()[,-1]))},spacing="xs")
output$contents52 <- renderTable({tail(data.frame(date=as.character(OtherClean()[,1]),OtherClean()[,-1]))},spacing="xs")
output$texto51<-renderText({paste("Number of uploaded rows =",as.character(nrow(OtherClean())))})
output$texto52<-renderText({paste("Number of deleted NA rows =",as.character(nrow(Other())-nrow(OtherClean())))})
#--- Data scaling & projection:
newdata<-eventReactive(input$go52,{as.matrix(scale(OtherClean()[,-1],center = SOMoutput()$CenterVal, scale = SOMoutput()$ScaleVal))})
Projection<-eventReactive(input$go52,{if (ncol(OtherClean())!=ncol(SOMoutput()$ExpClean)) {return(NULL)} else {map(SOMoutput()$som_model, newdata())}})
#--- Projection Outputs
BmusPROJ<-reactive({Projection()$unit.classif})
QerrsPROJ<-reactive({Projection()$distances})
############################### CORRETTO DA QUI: 20210211 ###############################################
BmusFreqPROJ<-reactive({factor(BmusPROJ(), levels=c(1:SOMoutput()$Dims$munits))})
TabHitsPROJ<-reactive({data.frame(table(BmusFreqPROJ()))})
HitsPROJ<-reactive({TabHitsPROJ()$Freq})
############################### A QUI. ######################################################################
output$texto53<-renderPrint({if (input$go52==0) {cat("", sep = "")
} else if (ncol(OtherClean())!=ncol(SOMoutput()$ExpClean)) {cat("Upload a file with ", ncol(SOMoutput()$ExpClean), " columns!", sep = "")
} else { cat("Projection of ", length(BmusPROJ()), " samples on", sep = "")
cat("\n SOM of size ", SOMoutput()$Dims$Row, "x", SOMoutput()$Dims$Col, sep = "")
cat("\n Topology: hexagonal",sep = "")
cat("\n Neighbourhood function: ",SOMoutput()$Neigh, sep = "")
cat("\n Distance measure used: Euclidean",sep = "")
cat("\n Number of training epochs: ",SOMoutput()$Epochs,sep = "")}
})
#---- Cluster assignment (respect to Kmeans graphs in third tab!)
############################### CORRETTO DA QUI: 20210211 ###############################################
ClusProj<-function() {D<-rep(0,length(BmusPROJ()))
for (i in c(2:length(KmeansForPlot()$centroids))) {d<-BmusClus(BmusPROJ(),KmeansForPlot()$clusNum[,i]);
D<-cbind(D,d)}; return(data.frame(D))}
############################### A QUI. ######################################################################
#--- Save Projection Outputs
OutputPdata<-reactive({list(OtherClean=OtherClean(),newdata=newdata(),HitsPROJ=HitsPROJ(),BmusPROJ=BmusPROJ(),QerrsPROJ=QerrsPROJ(),ClusProj=ClusProj())})
output$down51<-downloadHandler(filename=function() {paste0("Projection_output_",format(Sys.time(),format="%Y-%m-%d_%H-%M"),".RData")},
content = function(file) {
SOMoutput<-OutputPdata()
list.save(SOMoutput,file=file)}
)
#-- If ProjectionOutput file is loaded:
ProjectionOutput <- eventReactive(input$go53,{list.load(input$OtherOutput$datapath)})
ProjForPlot<-reactive({if(input$check52==FALSE){OutputPdata()} else {ProjectionOutput()} })
output$texto54<-renderPrint({if(input$go53==0){cat("", sep = "")} else {cat("Projection of ", length(ProjForPlot()$BmusPROJ), " samples loaded", sep = "")}
})
#-- SOM map features (Projection)
output$UIinput51 <- renderUI({if(input$menu51==" ") {
tagList(div(style= 'font-size:1em;',selectInput("menu52", label="Plot type", c(" "), selected = " ",width="80%")))
} else if (input$menu51=="hits" | input$menu51=="qerrs"){tagList(div(style= 'font-size:1em;',
selectInput("menu52", label="Plot type", c("grayscale","black filling"), selected = "grayscale",width="80%")))}
})
Switch5<-reactive({paste0(input$menu51,input$menu52)})
Plot51<- function() {switch(Switch5(),
" "={return(NULL)},
"hitsgrayscale"={SOMEnv::HexaHitsQuant(ProjForPlot()$HitsPROJ,SOMoutput()$Coord,SOMoutput()$Dims$Row,SOMoutput()$Dims$Col)},
"hitsblack filling"={SOMEnv::HexaHits(ProjForPlot()$HitsPROJ,SOMoutput()$Coord,SOMoutput()$Dims$Row,SOMoutput()$Dims$Col)},
"qerrsgrayscale"={SOMEnv::HexaQerrsQuant(ProjForPlot()$BmusPROJ,ProjForPlot()$QerrsPROJ,SOMoutput()$Coord,SOMoutput()$Dims$Row,SOMoutput()$Dims$Col)},
"qerrsblack filling"={SOMEnv::HexaQerrs(ProjForPlot()$BmusPROJ,ProjForPlot()$QerrsPROJ,SOMoutput()$Coord,SOMoutput()$Dims$Row,SOMoutput()$Dims$Col)})
}
output$plot51<- renderPlot({Plot51()})
output$down52 <- downloadHandler(filename =function() {paste0("SOM_projection_",input$menu51,"_",input$menu52,".pdf")},
content = function(file) {
pdf(file,height=input$h51,width=input$w51)
Plot51()
dev.off()}
)
#-- Basic statistics
Table51<-function() {SOMEnv::paramQuant(ProjForPlot()$HitsPROJ)}
output$down53 <- downloadHandler(filename ="Hits_projection_basic statistics.txt",
content = function(file) {
write.table(Table51(), file, row.names = F,col.names=T,sep="\t",quote=F)}
)
Table52<-function() {SOMEnv::paramQuant(ProjForPlot()$QerrsPROJ)}
output$down54 <- downloadHandler(filename ="Qerrs_projection_basic statistics.txt",
content = function(file) {
write.table(Table52(), file, row.names = F,col.names=T,sep="\t",quote=F)}
)
#-- Cluster profiles (projection) --> respect to fourth tab parameters!!!
############################### CORRETTO DA QUI: 20210211 ###############################################
ClusProj4<-reactive({SOMEnv::NClusChange(ProjForPlot()$ClusProj[,input$n43],Centroids4(),numSeq41())})
Plot52<- function() {if (is.null(ProjForPlot())) {return(NULL)
} else if (is.null(ProjForPlot())) {return(NULL)
} else if (length(numSeq41())!=input$n43) {return(NULL)
} else {switch(input$menu53,
"Proj by cluster"={SOMEnv::BoxUnits(ProjForPlot()$newdata,ClusProj4(),Centroids4(),Ylim=numSeq43(),pitch=numSeq42(),xdim=input$x41)},
"Proj by variable"={SOMEnv::BoxClus(c(input$n44,input$n45),ProjForPlot()$OtherClean[,-1],ClusProj4(),Centroids4())})}
}
############################### A QUI. ######################################################################
output$plot52<-renderPlot({if (is.null(ProjForPlot())) {return(NULL)
} else {Plot52()}
})
output$down55 <- downloadHandler(filename =function() {paste0("Cluster profiles_projection",".pdf")},
content = function(file) {
pdf(file,height=input$h51,width=input$w51)
Plot52()
dev.off()}
)
#---------------------------------------------------------------------------------------------------------------------------------------------------------------------
# server: SIXTH TAB
#---------------------------------------------------------------------------------------------------------------------------------------------------------------------
#--- Scelta dataset
ForDailyGraph<- function() {switch(input$menu61,
"training"={data.frame(date=SOMoutput()$ExpClean$date,Cluster=ClusExp4(),Bmus=SOMoutput()$Bmus,Qerrs=SOMoutput()$Qerrs)},
"projection"={data.frame(date=ProjForPlot()$OtherClean$date,Cluster=ClusProj4(),Bmus=ProjForPlot()$BmusPROJ,Qerrs=ProjForPlot()$QerrsPROJ)})
}
#--- Guess Obs/Day
Guess<-reactive({format(ForDailyGraph()$date,"%d/%m/%Y")})
Guess2<-reactive({ordered(Guess(), levels = unique(Guess()))})
Count<-reactive({plyr::count(Guess2())})
Total<-reactive({round(median(Count()$freq,na.rm=T),digit=0)})
output$UIinput63 <- renderUI({tagList(div(style= CSS32,numericInput("n61", label="Obs/Day", Total(),width="80%"))
)})
#--- Scelta intervallo dati
dateStart61<-function() {paste(
substr(ForDailyGraph()[1,"date"], 9, 10),"/",
substr(ForDailyGraph()[1,"date"], 6, 7),"/",
substr(ForDailyGraph()[1,"date"], 1, 4),sep="")}
dateEnd61<-function() {paste(
substr(ForDailyGraph()[nrow(ForDailyGraph()),"date"], 9, 10),"/",
substr(ForDailyGraph()[nrow(ForDailyGraph()),"date"], 6, 7),"/",
substr(ForDailyGraph()[nrow(ForDailyGraph()),"date"], 1, 4),sep="")}
output$UIinput61 <- renderUI({tagList(
div(style= CSS32,textInput(inputId = "text61",label = "Starting date:",value=dateStart61(),width="80%")),
div(style= CSS32,textInput(inputId = "text62",label = "Ending date:",value=dateEnd61(),width="80%"))
)})
Selection<-reactive({selectByDate(ForDailyGraph(),start=input$text61,end=input$text62)})
#--- Daily Graph
############################### CORRETTO DA QUI: 20210211 ###############################################
Plot61<- function() {SOMEnv::DailyBar(Selection(),Selection()$Cluster,Centroids4(),colSeq=colSeq4(),Total=input$n61,xdim=input$x61,ydim=0.8)}
############################### A QUI. ######################################################################
output$plot61<- renderPlot({Plot61()})
output$down61 <- downloadHandler(filename =function() {paste0("Daily graph",".pdf")},
content = function(file) {
pdf(file,height=input$h61,width=input$w61)
Plot61()
dev.off()}
)
#--- Download tables:
############################### CORRETTO DA QUI: 20210211 ###############################################
Table61<-function() {SOMEnv::FreqD(Selection()$date,Selection()$Cluster,Centroids4(),Total=input$n61)}
############################### A QUI. ######################################################################
output$down62 <- downloadHandler(filename ="Interval selection.txt",
content = function(file) {
write.table(Table61(), file, row.names = T,col.names=T,sep="\t",quote=F)}
)
############################### CORRETTO DA QUI: 20210211 ###############################################
Table62<-function() {switch(input$menu61,
"training"={SOMEnv::Freq(ClusExp4(),Centroids4())},
"projection"={SOMEnv::Freq(ClusProj4(),Centroids4())})
}
############################### A QUI. ######################################################################
output$down63 <- downloadHandler(filename ="Overall.txt",
content = function(file) {
write.table(Table62(), file, row.names = F,col.names=T,sep="\t",quote=F)}
)
############################### CORRETTO DA QUI: 20210211 ###############################################
Table63<-function() {SOMEnv::FreqM(ForDailyGraph()$date,ForDailyGraph()$Cluster,Centroids4())}
############################### A QUI. ######################################################################
output$down64 <- downloadHandler(filename ="Monthly.txt",
content = function(file) {
write.table(Table63(), file, row.names = F,col.names=T,sep="\t",quote=F)}
)
############################### CORRETTO DA QUI: 20210211 ###############################################
Table64<-function() {data.frame(Unit=c(1:length(clusNum4())),Cluster=clusNum4())}
############################### A QUI. ######################################################################
output$down66 <- downloadHandler(filename ="Unit cluster assignment.txt",
content = function(file) {
write.table(Table64(), file, row.names = F,col.names=T,sep="\t",quote=F)}
)
Table65<-function() {data.frame(ForDailyGraph())}
output$down67 <- downloadHandler(filename =function() {paste0(as.character(input$menu61)," data cluster assignment.txt")},
content = function(file) {
write.table(Table65(), file, row.names = F,col.names=T,sep="\t",quote=F)}
)
#--- Single sample profile
output$UIinput62 <- renderUI({tagList(
div(style= CSS32,textInput(inputId = "text63",label = "Select sample:",value=as.character(ForDailyGraph()[2,"date"]),width="100%"))
)})
#--- Starting plot
SelDay<-reactive({ForDailyGraph()[which(as.character(ForDailyGraph()$date)==input$text63),]})
output$texto61<-renderPrint({ cat(" BMU = ", SelDay()[,"Bmus"],sep = "")
cat("\n Cluster = ", SelDay()[,"Cluster"], sep = "")
cat("\n Qe = ", round(SelDay()[,"Qerrs"],digit=3), sep = "")
})
SelExpClean<- function() {switch(input$menu61,
"training"={data.frame(date=SOMoutput()$ExpClean$date,SOMoutput()$dataset)},
"projection"={data.frame(date=ProjForPlot()$OtherClean$date,ProjForPlot()$newdata)})
}
SelBOX<-reactive({SelExpClean()[which(as.character(SelExpClean()$date)==input$text63),]})
SingleSampPlot<- function() {
SLN<-SelDay()[,3]
CLS<-SelDay()[,2]
CexLy<-0.9
CexAxy<-0.85
Ylab<-c("Cluster")
Ymin<-0.5
Ymax<-input$n43+0.5
nClus<-input$n43
Names<-colnames(SOMoutput()$codebook)
############################### CORRETTO DA QUI: 20210211 ###############################################
Xlim<-c(0.5,ncol(SOMoutput()$codebook)+0.5)
############################### A QUI. ######################################################################
Ylim<-numSeq43()
pitch<-numSeq42()
Xdim<-input$x41
i<-1
#---qua il primo grafico:
opar <- par(fig=c(0, 0.5, 0, 1),oma=c(1,1,1,1),mar=c(1,1.2,1,1.2),xpd=TRUE,pty="m",family="serif");
SOMEnv::HexagonsClus(Centroids4(), clusNum4(), BCentr4(),
SOMoutput()$Coord, SOMoutput()$Dims$Row, SOMoutput()$Dims$Col, colSeq = colSeq4())
mtext(SelDay()[,"date"][i],side=3,lin=0.5);
#--qua ultimo dato:
points(SOMoutput()$Coord$X[SLN[i]], SOMoutput()$Coord$Y[SLN[i]],pch=16,cex=1,col="black")
mtext(paste0("Qe=", round(SelDay()[,4][i],digit=3)),side=1,lin=0.5);
#--------qua parte destra con boxplot:
#-------------------senza whiskers!!!!--------------------------------
StepB<-1/nClus;
SeqB<-abs(seq(-1,0,StepB));
NULLO<-data.frame(matrix(-100000,nrow=nClus,ncol=ncol(SOMoutput()$codebook)))
NULLO[SelDay()[i,"Cluster"],]<-SelBOX()[i,-1]
for (w in c(1:nClus)) {
############################### CORRETTO DA QUI: 20210211 ###############################################
opar <- par(fig=c(0.5, 1, SeqB[w+1], SeqB[w]),mar=c(1,1,1,0.5),oma=c(0.5,0.5,0,0.5),xpd=FALSE,pty="m",family="serif");
############################### A QUI. ######################################################################
opar <- par(new=TRUE);
boxplot(SOMoutput()$codebook[which(clusNum4()==w),1],range=0,xaxt="n",yaxt="n",xlim=Xlim,ylim=Ylim,boxwex=0.9,at=1,whisklty = 0, staplelty = 0);
abline(h=pitch,col="gray");
for (f in c(1:ncol(SOMoutput()$codebook)))
{boxplot(SOMoutput()$codebook[which(clusNum4()==w),f],range=0,xaxt="n",yaxt="n",boxwex=1,at=f,add=TRUE,whisklty = 0, staplelty = 0)};
axis(1,at=seq(1,ncol(SOMoutput()$codebook),1),labels=NA,tcl=-0.3,cex.axis=Xdim);
############################### CORRETTO DA QUI: 20210211 ###############################################
axis(1,at=seq(1,ncol(SOMoutput()$codebook),1),labels=Names,lwd=0,line=-0.5,cex.axis=Xdim);
############################### A QUI. ######################################################################
axis(2,at=pitch,labels=NA,tcl=-0.3,cex.axis=0.6);
############################### CORRETTO DA QUI: 20210211 ###############################################
axis(2,at=pitch,labels=pitch,lwd=0,line=-0.7,cex.axis=0.6,las=2);
points(seq(1,ncol(SOMoutput()$codebook),1),NULLO[w,], pch=16,col="red",cex=0.7)
############################### A QUI. ######################################################################
on.exit(par(opar))
}
on.exit(par(opar))
} # END SingleSampPlot
output$plot62<- renderPlot({SingleSampPlot()})
output$down65 <- downloadHandler(filename =function() {paste0("Single sample",".pdf")},
content = function(file) {
pdf(file,height=input$h61,width=input$w61)
SingleSampPlot()
dev.off()}
)
#------------------------------------------------------------------------------------------------
session$onSessionEnded(function() {
stopApp()
})
} # server END
#---------------------------SHINY APP BUILT UP -------------------------------------
shinyApp(ui, server)
Try the SOMEnv package in your browser
Any scripts or data that you put into this service are public.
SOMEnv documentation built on July 26, 2021, 5:06 p.m.
R Package Documentation
Browse R Packages
We want your feedback!
Note that we can't provide technical support on individual packages. You should contact the package authors for that.
## App corrected 20210211
requiredPackages <- c("dplyr","plyr","openair", "rlist", "shiny", "kohonen", "shinycssloaders", "shinycustomloader")
ipak <- function(pkg){
new.pkg <- pkg[!(pkg %in% installed.packages()[, "Package"])]
if (length(new.pkg))
install.packages(new.pkg, dependencies = TRUE)
sapply(pkg, require, character.only = TRUE)
}
ipak(requiredPackages)
#---- Upload file fino a 100 MB consentito:
options(shiny.maxRequestSize = 100*1024^2) # per consentire upload files fino a 100 MB
#--------------------------- CSS STYLE -----------------------------------------
CSSHeader<-'font-family: Tahoma, Geneva, sans-serif;
font-size: 22px;
letter-spacing: 2px;
word-spacing: 2px;
color: #000000;
font-weight: 400;
text-decoration: none;
font-style: normal;
font-variant: normal;
text-transform: none'
CSS11<-"[type = 'number'] {font-size:1em; height:25px}
[type = 'text'] {font-size:1em; height:25px}
# [type = 'label'] { font-size:80%}
"
CSS12<-'font-size:1em'
CSS13<-'padding-left: 1px;padding-right: 1px;padding-top: 1px;padding-bottom: 1px;margin-top: -10px'
CSS14<-'height:1em;font-size:1em'
CSS15<-'padding-left: 1px;padding-right: 1px'
CSS16<-'font-size:1em;overflow-x: scroll'
#------------------------------------------------------
CSS21<-"[type = 'number'] {font-size:1em; height:25px}
[type = 'text'] {font-size:1em; height:25px}
# [type = 'label'] { font-size:80%}
"
CSS22<-'padding-left: 1px;padding-right: 1px;padding-top: 5px;padding-bottom: 1px;margin-top: 1px'
CSS23<-'font-size:14px'
CSS24<-'height:11px;font-size:1em'
CSS25<-'padding-left: 1px;padding-right: 1px;padding-top: 1px;'
CSS26<-'font-size:1em;overflow-x: auto;overflow-y: auto'
#------------------------------------------------------
CSS31<-"[type = 'number'] {font-size:1em; height:25px}
[type = 'text'] {font-size:1em; height:25px}
# [type = 'label'] { font-size:80%}
"
CSS32<-'font-size:1em'
CSS33<-'align:left;vertical-align: middle'
CSS331<-'font-size:1em;vertical-align: middle;align: center'
CSS34<-'font-size:11px;text-align: center'
CSS35<-'overflow-y: scroll'
#------------------------------------------------------
CSS41<-'overflow-x: scroll;overflow-y: scroll'
CSS42<-'font-size:1em;overflow-x: scroll;overflow-y: scroll'
#------------------------------------------------------
CSS51<-'padding-left: 1px;padding-right: 1px;padding-top: 1px;padding-bottom: 1px;margin-top: 1px'
##########################################################################################################################################################
#
# PAGE DEFINITION
#
##########################################################################################################################################################
ui <- pageWithSidebar(
headerPanel(div(style=CSSHeader,"SOMEnv Graphical User Interface (v 1.1.2)"),windowTitle = "SOMEnv GUI"),
sidebarPanel(width=3,style='background: white; padding: 1px',
fluidRow(style='height:10vh;background: white;padding-top: 1px;padding-bottom: 1px;padding-left: 1px;padding-right: 1px',
img(src='Logo.jpg', align = "center")#---------------------------------------------------------------------------------------- DA RIVEDERE
),# fluidRow END
fluidRow(style='overflow-y:scroll;height:20vh;background: pink;padding-top: 1px;padding-bottom: 1px;padding-left: 1px;padding-right: 1px',
tags$h4("An R package for mining information from multivariate environmental high frequency data by Self-Organizing Map and k-means clustering algorithms"),
tags$h5("Authors: S. Licen, M. Franzon, T. Rodani, P. Barbieri"),
tags$h5("Licence: GPL-3.0"),
tags$h5("Large datasets up to 100 MB are allowed!")
),# fluidRow END
fluidRow(style='overflow-y:scroll;height:60vh;background: mistyrose;padding-top: 5px;padding-bottom: 1px;padding-left: 1px;padding-right: 1px',
h4("Help"),
br(),
tags$p("
", tags$strong("Tab 1: Load Data")," ", tags$br() ,"
Use Browse button to select the data file. Data are imported in the GUI using the ", tags$a(href=" https://www.rdocumentation.org/packages/openair/versions/2.7-4/topics/import","import"),"function from", tags$a(href="https://cran.r-project.org/web/packages/openair/index.html","openair"), "package.
The data must be in table format in a txt file with date variable (or datetime variable)
in the first column (the header must be 'date') and numeric variables in the following columns.
The user has to write the correct datetime format according to the instructions for", tags$a(href="https://cran.r-project.org/web/packages/openair/index.html", "strptime"),"function used in import function.
Column and decimal separator have to be selected as well. Press Load button to load the data.
When loaded, the first six rows (header) and the last six rows (tail) appear in the lower part of the screen thus the user can check
if the data have been uploaded correctly.
Under the above mentioned tables the number of total uploaded rows as well as the number of deleted rows (containing NA values) are presented.
If the date format is not correct and/or one or more columns contain non numeric values some warnings appear.
When the data are correctly loaded the user can move to the next tab.
"),
tags$p("
", tags$strong("Tab 2: SOM training")," ", tags$br() ,"
The SOM training is based on the use of",tags$a(href="https://www.rdocumentation.org/packages/kohonen/versions/3.0.10/topics/supersom", "som"), "function from", tags$a(href="https://cran.r-project.org/web/packages/kohonen/index.html", "kohonen package"), ". The parameter selection for training the SOM has not default values but some heuristic rules by Vesanto et al. (2000) are widely used, these rules have been embedded in the GUI. Thus, once the data are uploaded, the parameters related to that specific data are calculated. The map dimensions are calculated with three possible specifications (“small”, “medium”, ”large”), and the number of epochs and neighborhood radius vary accordingly. Nevertheless the user can choose custom values for every variable for training the SOM. The neighborhood function can be chosen between “gaussian” and “bubble”. For more details about parameter selection see Clark et al. (2020).
Before training the map the data are by default scaled by variable to have zero mean and variance 1. Scaling is used to ensure all variables have the same importance during training, regardless of absolute magnitude. Moreover an initialization matrix for the SOM units is calculated based on the first two eigenvectors of the data as in Vesanto et al. (2000). Three parameters used in the som function from kohonen package are set by default: topology (“hexagonal”), distance measure (“Euclidean”), learning algorithm (“pbatch”).
When the parameters are chosen pressing the Train SOM button starts the training.
When the calculation is finished in the lower part of the screen on the left the first six rows of the normalized codebook (unit vectors of modeled variables) appear as well as a summary of the training parameters (on the right).
Then the user must save the results (a .RData file) using the button Download SOM output.
For details about the file content see the Appendix below.
If needed, the user can evaluate the quantization error (QE) and topographic error (TE) for the trained map using the Evaluate QE and Evaluate TE buttons respectively.
For more information about these parameters see Clark et al. (2020). Afterwards the user can move to the next tab.
"),
tags$p("
", tags$strong("Tab 3: SOM Map")," ", tags$br() ,"
This tab is provided to visualize the results obtained with the SOM training.
First a “SOM_output” .RData file (obtained from the previous tab) has to be loaded using Browse and Load buttons.
When loaded, a summary of the training parameters used is shown on the left. In the lower part of the screen several plots can be visualized, namely:
Heatmaps: on the left part using Plot button the heatmaps of the SOM map are visualized (i.e. the values distribution on the map of each modeled variable). The filling of the hexagons represents the basic statistics (lower outliers = white, quartiles = gray scale, upper outliers = black) of each modeled variable. For the quartiles and outliers evaluation the", tags$a(href="https://www.rdocumentation.org/packages/graphics/versions/3.6.2/topics/boxplot", "boxplot"), "R function is used with default parameters. The grid (rows x columns) in which plotting the heatmaps can be tailored by the user (see controls above the plot). The download table button allows to download the basic statistics for the modeled variables.
SOM map features:"),
tags$ul(
tags$li("umat: the unified distance matrix (U-matrix) represents the distances between the units. It can be represented as green-white scale (“gs”) or grayscale (“bw”). Darker colors represent lower distances. The choice is related to the fact that the U-matrix can be seen as a representation of valleys (green) and mountains (white – snow). The presence of “mountains” can show the separation of clusters of data. For details see Ultsch et al. (1990);"),
tags$li("hits: this graph represents the distribution of the experimental data on the map, as each unit represents one or more experimental data. It can be visualized either as “grayscale” or “blackfilling”. In the latter the filling of the hexagons is proportional to the number of hits. For further details see also Licen et al. (2020a);"),
tags$li("qerrs: this graph represents the distribution of the quantization errors (QEs) of experimental data on the map, the QE accounts for the accuracy of the match of each experimental vector to the corresponding unit to which it is associated (Best Matching Unit ). It can be visualized either as “grayscale” or “blackfilling”. In the latter the filling of the hexagons is proportional to the quantization error relative values. For further details see also Licen et al. (2019).")
),
tags$p("
The download Hits bs and download Qerrs bs buttons allow to download tables containing the basic statistics for each feature respectively.
From this Tab on all the figures produced in the Tabs can be saved either “as is” using the right click button of the mouse (image format) or using the download pdf button in proximity of the figure. The figure dimensions can be chosen selecting Height and Width (in inches) present nearby the figure or, generally, in the Tab.
"),
tags$p("
", tags$strong("Tab 4: Kmeans clustering")," ", tags$br() ,"
In this Tab the user can further clusterize the data operating a second level clusterization on the codebook using a kmeans clustering algorithm (see Vesanto et al. (1999) and Clark et al. (2020) for details). The calculation are operated on the SOM_output .RData uploaded in the previous tab.
In brief for each number of clusters selected (k = from 2 to max) the kmeans clustering algorithm is run for 100 epochs obtaining a splitting of the units and a corresponding total quantization error for the codebook. The above mentioned algorithm can be run multiple times (iterating times) for each k, and the best of these is selected based on sum of squared errors. For each iteration the function selects a new random seed to set the initialization parameters of the algorithm (see Vesanto et al. (2000)). To obtain reproducible results a seed can be set. In this case the algorithm is evaluated only once (one iteration) for each k. Setting a seed results in fixing the random procedure in order to obtain reproducible. Once selected the max number of clusters and the number of iterating times, press the Run button to start the calculation. The evaluation can take several minutes depending on the number of units and experimental variables. Two progress bars are visualized in the lower right corner of the screen to check the progress of the calculation. At the end of the calculation a Davies-Bouldin index plot appears (see Davies-Bouldin (1979)). The best cluster split is in correspondence to the lower value of the index. Then the user must save the results (a .RData file) using the button Download kmeans output. For details about the file content see Appendix below. The N.clusters control displays automatically the best number of clusters according to Davies-Bouldin index, nevertheless the user can also explore the results for a different number of k. Using the Plot button the SOM map with the cluster split is visualized. The cluster numbers are positioned at the units which are the centroids of the clusters respectively. The user can change the cluster colors clicking on the legend rectangles.
The Cluster profiles graph shows respectively:"),
tags$ul(
tags$li("by cluster: modeled experimental variables profiles of units split by clusters represented by box plots. The boxes show the interquartile range, the thick line shows the median (box whiskers are omitted for the sake of the readability of the figure). The normalized values are used to fit all the variables in the same graph. The user can change some graphical parameters using the controls under the graph;"),
tags$li("by variable: each graph shows an experimental variable, as modeled by the SOM algorithm, split by cluster and depicted by boxplots (using the boxplot R function with default parameters). The values are denormalized to allow checking how the experimental variables were modeled. The grid (rows x columns) in which plotting the heatmaps can be tailored by the user (see controls under the plot);"),
tags$li("Exp by cluster: variables profiles of experimental data split by clusters represented by box plots. The boxes show the interquartile range, the thick line shows the median (box whiskers are omitted for the sake of the readability of the figure). The normalized values are used to fit all the variables in the same graph. The user can change some graphical parameters using the controls under the graph;"),
tags$li("Exp by variable: each graph shows an experimental variable, as present in the experimental data, split by cluster and depicted by boxplots (using the boxplot R function with default parameters). The values are denormalized. The grid (rows x columns) in which plotting the heatmaps can be tailored by the user (see controls under the plot).")
),
tags$p("
See Licen et al. (2018) and Licen et al. (2020b) for details.
"),
tags$p("
As the cluster numbering is random, if the user needs a different numbering of the clusters, the control named Cluster numbers: (upper right of the screen) can be used. The number of clusters has to be expressed by a comma delimited sequence of numbers without spaces. Example: if the default numbering is 1,2,3 and the user needs that cluster 1 becomes 2, 2 -> 3 and 3 ->1, the sequence to be written is 2,3,1. The numbering change affects only the plots, it is not saved in the .RData file. To save the custom cluster numbering see Tab 6.
The user can also load a previously obtained Kmeans_output.RData file (providing to previously load the corresponding SOM_output.RData file in Tab 3!!!) to explore the results. This is done checking the Load file? checkbox and loading the selected file. If the checkbox is unchecked again the user can go back to the latest kmeans evaluation.
"),
tags$p("
", tags$strong("Tab 5: Projection")," ", tags$br() ,"
This tab allows to project into the model data which have not been used to build it. The data to be projected must have the same number of columns (variables) as the training data with date variable (or datetime variable) in the first column (the header must be 'date'). The data file is chosen using the Browse button. Press the Load button to load the data. When loaded, the first six rows (header) and the last six rows (tail) appear in the upper right part of the screen thus the user can check if the data have been uploaded correctly. The number of total uploaded rows as well as the number of deleted rows (containing NA values) are presented. Pressing the Projection button the data are normalized with the same parameters as the training data and projected into the model (referred to the .RData files loaded in Tab 3 and Tab 4). Then the user must save the results (a .RData file) using the button named Download Projection. For details about the file content see Appendix below. Results can be explored visualizing them on the SOM map, for the options description (hits and qerrs) refer to Tab 3. The cluster profiles are built as explained for Tab 4 (options Exp by cluster and Exp by variable) but using the projected data. The visualization parameters of the latter are duplicated from those selected in Tab 4.
The user can also load a previously obtained 'Projection_output' .RData file (providing to previously load the corresponding SOM_output .RData file in Tab 3 and Kmeans_output in Tab 4!!!) to explore the results. This is done checking the Load file? checkbox and loading the selected file. If the checkbox is unchecked again the user can go back to the latest projection.
"),
tags$p("
", tags$strong("Tab 6: Daily profiles")," ", tags$br() ,"
The results obtained in the previous tabs can be explored considering their time sequence. In this tab either the results obtained with the training data (from Tab 3) or the projected data (Tab 5) can be visualized. The visualization parameters of the graphs (e.g. colors, cluster numbering, Y range for the cluster profiles, etc..) are duplicated from those selected in Tab 4. The user can choose the time period to be represented using the controls Starting date and Ending date. The user has to check the control Obs/Day which shows the number of observations for each day as guessed by the function. If the number is not correct, it has to be changed accordingly. Example: if the observations (samples) were recorded every 15 min then the number to be put in the control has to be 4 x 24 hours = 96.
The Daily graph shows the cluster percentage frequency for each day (the X labels dimensions can be changed using the control above the graph). The corresponding table can be downloaded using the table button. The overall tab button downloads a table containing the cluster percentage splitting for the full period. The month tab button downloads a table containing the cluster percentage split by month.
On the upper right of the screen a single observation (sample) can be selected. The BMU number, cluster number and quantization error of the sample are shown. The figure below uses the SOM map and cluster profiles of modeled variables as a legend to show the positioning of the sample respect to the evaluated model: the BMU is labelled by a black dot on the SOM map and the (normalized) experimental values of the sample are shown by red dots onto the corresponding cluster profile. The figure can be useful to visualize possible outliers (see Licen et al. (2018) and Licen et al. (2019)). Two buttons allow to download a table containing the cluster number assignment to the map units (Units cluster assignment) and the to the data (Data cluster assignment, either training or projection as selected in control Dataset). The cluster numbering corresponds to that selected by the user in Tab 4 (Cluster numbers: control).
"),
tags$p("
", tags$strong("References")," ", tags$br() ,""),
tags$ul(
tags$li("Clark et al. (2020): Clark, S., Sisson, S.A., Sharma, A. Tools for enhancing the application of self-organizing maps in water resources research and engineering (2020) Adv. Water Resour. 143, art. no. 103676. DOI: 10.1016/j.advwatres.2020.103676", tags$a(href="https://www.sciencedirect.com/science/article/pii/S030917081931139X", "Link")),
tags$li("Davies-Bouldin (1979): D.L. Davies, D.W. Bouldin, A cluster separation measure, IEEE Trans. Pattern Anal. Mach. Intell. 1 (2) (1979) 224–227, http://dx.doi.org/10.1109/TPAMI. 1979.4766909.", tags$a(href="https://ieeexplore.ieee.org/document/4766909","Link"),""),
tags$li("Licen et al. (2018): Licen, S., Barbieri, G., Fabbris, A., Briguglio, S.C., Pillon, A., Stel, F., Barbieri, P. Odor control map: Self organizing map built from electronic nose signals and integrated by different instrumental and sensorial data to obtain an assessment tool for real environmental scenarios (2018) Sens Actuators, B Chem, 263, pp. 476-485. DOI: 10.1016/j.snb.2018.02.144", tags$a(href="https://www.sciencedirect.com/science/article/pii/S0925400518304246","Link"),""),
tags$li("Licen et al. (2019): Licen, S., Cozzutto, S., Barbieri, G., Crosera, M., Adami, G., Barbieri, P. Characterization of variability of air particulate matter size profiles recorded by optical particle counters near a complex emissive source by use of Self-Organizing Map algorithm (2019) Chemometr. Intelligent Lab. Syst., 190, pp. 48-54. DOI: 10.1016/j.chemolab.2019.05.008", tags$a(href="https://www.sciencedirect.com/science/article/pii/S0169743919301376", "Link"),""),
tags$li("Licen et al. (2020a): Licen, S., Cozzutto, S., Barbieri, P. Assessment and comparison of multi-annual size profiles of particulate matter monitored at an urban-industrial site by an optical particle counter with a chemometric approach (2020) Aerosol Air Qual. Res., 20 (4), pp. 800-809.DOI: 10.4209/aaqr.2019.08.0414", tags$a(href="https://aaqr.org/articles/aaqr-19-08-oa-0414", "Link"),""),
tags$li("Licen et al. (2020b): Licen, S., Di Gilio, A., Palmisani, J., Petraccone, S., de Gennaro, G., Barbieri, P. Pattern recognition and anomaly detection by self-organizing maps in a multi month e-nose survey at an industrial site (2020) Sensors, 20 (7), art. no. 1887. DOI: 10.3390/s20071887.", tags$a(href=" https://www.mdpi.com/1424-8220/20/7/1887","Link"),""),
tags$li("Ultsch et al. (1990): A. Ultsch, H.P. Siemon, Kohonen’s self organizing feature maps for exploratory data analysis, in: Proceedings of International Neural Network Conference (INNC’90), Kluwer academic Publishers, Dordrecht, 1990, pp. 305–308."),
tags$li("Vesanto et al. (1999): Vesanto, J., 1999. SOM-based data visualization methods. Intelligent Data Analysis 3, 111–126. doi:10.1016/S1088-467X(99)00013-X", tags$a(href="https://www.sciencedirect.com/science/article/abs/pii/S1088467X9900013X?via%3Dihub", "Link"),""),
tags$li("Vesanto et al. (2000): J. Vesanto, J. Himberg, E. Alhoniemi, J. Parhankagas, SOM Toolbox for Matlab 5, Report A57, 2000, Available at: www.cis.hut.fi/projects/somtoolbox/package/papers/techrep.pdf", tags$a(href="www.cis.hut.fi/projects/somtoolbox/package/papers/techrep.pdf","Link"),"")
),
tags$p("
", tags$strong("Appendix")," ", tags$br() ,"
In this section the content of the output files is described. The files can be imported in R software using the function 'Load workspace' in the File menu. An object called 'x' is loaded. The content is detailed in the following:
SOM_output.RData "),
tags$ul(
tags$li("$ExpClean: experimental data as imported by the software;"),
tags$li("$dataset: normalized experimental data;"),
tags$li("$ScaleVal: values of column scaling;"),
tags$li("$CenterVal: values of column centering;"),
tags$li("$Epochs: number of training epochs;"),
tags$li("$Neigh: neighborhood type;"),
tags$li("$Dims: map dimensions ($Row and $Col);"),
tags$li("$Coord: coordinates of hexagon centers in the map;"),
tags$li("$codebook: prototype vectors;"),
tags$li("$DeCod: denormalized codebook (variables in original units);"),
tags$li("$Umat: Unified Distance Matrix;"),
tags$li("$Hits: number of hits for each prototype;"),
tags$li("$Bmus: Best Matching Unit (BMU) for each experimental vector;"),
tags$li("$Qerrs: quantization error for each experimental vector."),
),
tags$p("
Kmeans_output.RData:
"),
tags$ul(
tags$li("$centroids: a list containing the centroids for each of the centroids number evaluated (k) ( e.g. $centroids[[5]] is a matrix containing the centroids for five clusters);"),
tags$li("$clusNum: dataframe containing the cluster assignment for prototypes (column 2 for two clusters, etc...);"),
tags$li("$ind: a vector containing the DB-index value for each cluster;"),
tags$li("$err: a vector containing the error for each cluster;"),
tags$li("$BCentr: a list containing the BMU for each centroid ( e.g. $BCentr [[5]] is a vector containing the BMU of centroids for five clusters);"),
tags$li("$ClusExp: dataframe containing the cluster assignment for experimental vectors (column 2 for two clusters, etc...);"),
tags$li("$Seed: the seed number used to start the kmeans algorithm and obtain the corresponding solution."),
),
tags$p("
Projection_output.RData:
"),
tags$ul(
tags$li("$OtherClean: projected data as imported by the software;"),
tags$li("$newdata: normalized projected data normalized with the same parameters as experimental data;"),
tags$li("$HitsPROJ: number of hits for each prototype related to projected data;"),
tags$li("$BmusPROJ: Best Matching Unit (BMU) for each projected vector;"),
tags$li("$QerrsPROJ: quantization error for each projected vector;"),
tags$li("$ClusProj: dataframe containing the cluster assignment for projected vectors (column 2 for two clusters, etc...).")
)
)# fluidRow END
),# END sidebarPanel
mainPanel(width=9,style='background: white',
tabsetPanel(
#---------------------------------------------------------------------------------------------------------------------------------------------------------------------
# FIRST TAB
#---------------------------------------------------------------------------------------------------------------------------------------------------------------------
tabPanel("Load data", fluid = TRUE,
# tags$style(CSS21),
fluidRow(width='80%', style='background: aliceblue;',
# Input: File input and help ----
column(5,tags$div(style= 'font-size:1em',fileInput("Experimental", label="Load experimental data:", accept=c("txt"),buttonLabel=div(style= CSS24,"Browse..."),width="100%")),
actionButton(inputId = "go11",label = "Load")
),# column END
column(3,
tags$div(style= 'font-size:1em',textInput("text11", label="Date format:","%Y-%m-%d %H:%M:%S",width="100%")),
tags$div(style='font-size:1em; font-style:italic; align: left',"Variable date must be in the first column"),
tags$div(style='font-size:1em; font-style:italic; align: left',"See openair import function for date format input")
),# column END
############################### CAMBIATO DA QUI: ############################################################
column(2,
tags$div(style= 'font-size:1em',textInput("text12", label="Separator:","\t",width="100%")),
tags$div(style= 'font-size:1em',textInput("text13", label="Decimal:",".",width="100%"))
############################### A QUI. ######################################################################
)# column END
),# fluidRow END
fluidRow(style='height:60vh;background: white;padding-top: 1px;padding-bottom: 1px;padding-left: 1px;padding-right: 1px',
column(5,verticalLayout(tags$h4("Experimental data header"),
tags$div(style= CSS26,tableOutput("contents11")),
tags$div(style= 'font-weight: bold',br()),
tags$div(style= 'font-weight: bold',textOutput("texto11")),
tags$div(style= 'font-weight: bold',textOutput("texto12")))),
column(5,verticalLayout(tags$h4("Experimental data tail"),
tags$div(style= CSS26,tableOutput("contents12")),
tags$div(style= 'font-weight: bold; color:red',textOutput("texto13"))))
)# fluidRow END
), # tabPanel END
#---------------------------------------------------------------------------------------------------------------------------------------------------------------------
# SECOND TAB
#---------------------------------------------------------------------------------------------------------------------------------------------------------------------
tabPanel("SOM training", fluid = TRUE,
# tags$style(CSS21),
fluidRow(style='height:20vh;background: aliceblue;',
tags$style(type='text/css', ".selectize-input { padding: 1px; min-height: 0;} .selectize-dropdown { line-height: 20px; }"),
# Input: SOM map dimensions ----
uiOutput("UIinput21"),
column(2,tags$div(style= 'font-size:1em',selectInput("Size", "Map size:", c("regular","small","big"), selected = "small",width="100%")),
tags$div(style= 'font-size:1em;font-weight: bold',textOutput("texto21"))),
# Input: SOM training parameters ----
uiOutput("UIinput22"),
# Input: SOM training parameters ----
column(2,tags$div(style= 'font-size:1em;',selectInput("Neigh", "Neighborhood:", c("bubble","gaussian"), selected = "gaussian",width="100%")),
uiOutput("UIinput23")),
column(2,actionButton(inputId = "go21",label = "Train SOM"))
),# fluidRow END
fluidRow(style='height:60vh;background: white;',
############################### CAMBIATO DA QUI: ############################################################
column(5,verticalLayout(h4("Codebook header"),
tags$div(style= CSS26,tableOutput("contents21")),
tags$h4("Map quality parameters"),
splitLayout(cellWidths=c("50%","50%"),actionButton(inputId = "go22",label = "Evaluate QE"),
tags$div(style= 'font-size:1em;',verbatimTextOutput("texto23",placeholder=TRUE))),
splitLayout(cellWidths=c("50%","50%"),actionButton(inputId = "go23",label = "Evaluate TE"),
tags$div(style= 'font-size:1em;',verbatimTextOutput("texto24",placeholder=TRUE))),
div(style='font-size:1em; font-style:italic; color:black',"(The TE calculation can take several minutes)"))),
############################### A QUI #######################################################################
column(4,verticalLayout(h4("Summary"),
tags$div(style= 'font-size:1em;',verbatimTextOutput("texto22",placeholder=TRUE)),
downloadButton("down21","Download SOM output")))
)# fluidRow END
), # tabPanel END
#---------------------------------------------------------------------------------------------------------------------------------------------------------------------
# THIRD TAB
#---------------------------------------------------------------------------------------------------------------------------------------------------------------------
tabPanel("SOM Map", fluid = TRUE,
# tags$style(CSS31),
tags$style(type='text/css', ".butt { padding-left: 2px; padding-right: 2px; padding-top: 1px; padding-bottom: 1px; min-height: 0;font-size:1em}"),
fluidRow(style='background: aliceblue;',
# Input: File input and help ----
column(5,tags$div(style= 'font-size:1em',fileInput("SOMoutput", label="Load SOM output:", accept=c("RData"),buttonLabel=div(style= CSS24,"Browse..."),width="100%")),
actionButton(inputId = "go31",label = "Load")
),# column END
column(5,splitLayout(cellWidths = c("20%","80%"),
tags$h4("Summary:"),
tags$div(style= 'font-size:1em;',verbatimTextOutput("texto31",placeholder=TRUE))))
),# fluidRow END
fluidRow(style='overflow-x:scroll;background:white;',
# Graphs: SOM map and cluster profiles ----
column(7,verticalLayout(splitLayout(cellWidths = c("20%","10%","10%","50%","10%"),
tags$h4("Heatmaps"),
actionButton(inputId = "go32",label = "Plot"),
tags$h4(""),
uiOutput("UIinput31"),
tags$h4("")),
tags$div(withSpinner(plotOutput("plot31"),type=6,color="green")),
splitLayout(cellWidths = c("10%","10%","5%","10%","10%","5%","25%","25%"),
tags$div(style= 'font-size:1em; font-weight: bold',"Height(in):"),
tags$div(style= CSS32,numericInput("h31", label=NULL, 5,width="100%")),
tags$h4(""),
tags$div(style= 'font-size:1em; font-weight: bold',"Width(in):"),
tags$div(style= CSS32,numericInput("w31", label=NULL, 7,width="100%")),
tags$h4(""),
downloadButton("down30","download table",class="butt"),
downloadButton("down31","download pdf",class="butt")
)
)),
column(3,verticalLayout(h4("SOM map features"),
tags$div(withSpinner(plotOutput("plot32"),type=6,color="green"))
)),
column(2,verticalLayout(h5("__________________"),
tags$div(style= 'font-size:1em;',selectInput("menu31", label="Feature", c(" ","umat","hits","qerrs"), selected = " ",width="80%")),
uiOutput("UIinput32"),
tags$div(style= 'font-size:1em; font-weight: bold',numericInput("h32", "Height(in):", 3,width="80%")),
tags$div(style= 'font-size:1em; font-weight: bold',numericInput("w32", "Width(in):", 2,width="80%")),
downloadButton("down32","download pdf",class="butt"),
tags$h5(""),
downloadButton("down33","download Hits bs",class="butt"),
tags$h5(""),
downloadButton("down34","download Qerrs bs",class="butt")
))
)# fluidRow END
), # tabPanel END
#---------------------------------------------------------------------------------------------------------------------------------------------------------------------
# FOURTH TAB
#---------------------------------------------------------------------------------------------------------------------------------------------------------------------
tabPanel("Kmeans clustering", fluid = TRUE,
# tags$style(CSS31),
fluidRow(style='background: aliceblue;',
# Input: File input and help ----
column(3,
tags$div(style= 'font-size:1em; font-weight: bold',numericInput("n41", "Max clusters:", 8,min=2,width="80%")),
tags$div(style= 'font-size:1em; font-weight: bold',numericInput("n42", "Iterating times (x 100 epochs):", 5,min=1,width="80%")) ,
splitLayout(cellWidths = c("30%","70%"),
actionButton(inputId = "go41",label = "Run"),
downloadButton("down41","Download Kmeans output",class="butt"))
),# column END
column(8,verticalLayout(splitLayout(cellWidths = c("15%","55%","10%","20%"),
checkboxInput(inputId = "check41",label = "Load file?",value=FALSE),
conditionalPanel(condition = "input.check41 == 1",
tags$div(style= 'font-size:1em',fileInput("KmeansOutput", label="Load Kmeans output:", accept=c("RData"),buttonLabel=div(style= CSS24,"Browse..."),width="90%"))),
conditionalPanel(condition = "input.check41 == 1",
tags$div(style=CSS33,actionButton(inputId = "go42",label = div(style=CSS331,"Load")))),
uiOutput("UIinput41")), # splitLayout END
tags$div(style='font-size:13px; font-style:italic; color:red',"(Remember to load SOM output in SOM map tab!)"),
uiOutput("UIinput42")
)# verticalLayout END
)# column END
),# fluidRow END
fluidRow(style='width:100%; overscroll-x:auto; background: white;',
# Input: File input and help ----
column(style='background: aliceblue',3,verticalLayout(
splitLayout(cellWidths = c("65%","35%"),
tags$div(style= 'font-size:11px',radioButtons(inputId = "choice41",label = "Set seed?",choices=c("Yes","No"),selected="No",inline=TRUE,width="90%")),
conditionalPanel(condition = "input.choice41 == 'Yes'",
tags$div(style= 'font-size:11px',numericInput("Seed", "", value=7,width="100%")))),
tags$h4("DB-index plot"),plotOutput("plot41"))
),# column END
column(3,verticalLayout(tags$h4("SOM map"),
withSpinner(plotOutput("plot42"),type=5,color="gray")
)),
column(6,verticalLayout(fluidRow(column(5,h4("Cluster profiles")),
column(5,div(style= 'font-size:1em',selectInput("menu41", label=NULL, c("by cluster","by variable","Exp by cluster","Exp by variable"), selected = "by cluster",width="90%")))),
withLoader(plotOutput("plot43"),type="text",loader=list(marquee("Calculating...",
direction="down", behavior = "scroll",width = "90%",style="font-size:15px; font-weight: bold; color:gray")))
))
),# fluidRow END
fluidRow(style='height:5vh;background: white;',
# Input: File input and help ----
column(3,splitLayout(cellWidths = c("15%","25%","15%","25%","5%","15%"),
tags$div(style= 'font-size:1em; font-weight: bold',"H(in):"),
tags$div(style= CSS32,numericInput("h41", label=NULL, 5,width="90%")),
tags$div(style= 'font-size:1em; font-weight: bold',"W(in):"),
tags$div(style= CSS32,numericInput("w41", label=NULL, 7,width="90%")),
tags$h4(""),
downloadButton("down42","pdf",class="butt")
)
),
column(2,splitLayout(cellWidths = c("40%","50%","10%"),
tags$h4(""),
downloadButton("down43","pdf",class="butt"),
tags$h4(""))
),
uiOutput("UIinput43")
)# fluidRow END
), # tabPanel END
#---------------------------------------------------------------------------------------------------------------------------------------------------------------------
# FIFTH TAB
#---------------------------------------------------------------------------------------------------------------------------------------------------------------------
tabPanel("Projection", fluid = TRUE,
# tags$style(CSS31),
fluidRow(style='height:30vh;background: aliceblue;',
# Input: File input and help ----
column(3,div(style= 'font-size:1em',fileInput("Other", label="Load dataset:", accept=c("txt"),buttonLabel=div(style= CSS24,"Browse..."),width="100%")),
fluidRow(column(4,actionButton(inputId = "go51",label = "Load")),
column(8,div(style='font-size:1em; font-style:italic; color:red',"(Remember to load SOM output in SOM map tab!)"))),
tags$div(style= 'font-size:1em;font-weight: bold',br()),
tags$div(style= 'font-size:1em;font-weight: bold',textOutput("texto51")),
tags$div(style= 'font-size:1em;font-weight: bold',textOutput("texto52"))
),# column END
column(2,
tags$div(style= 'font-size:1em',textInput("text51", label="Date format:","%Y-%m-%d %H:%M:%S",width="100%")),
tags$div(style='font-size:1em; font-style:italic; align: left',"Variable date must be in the first column"),
tags$div(style='font-size:1em; font-style:italic; align: left',"See openair import function for date format input")
),# column END
column(1,
############################### CAMBIATO DA QUI: ############################################################
tags$div(style= 'font-size:1em',textInput("text52", label="Separator:",",",width="100%")),
tags$div(style= 'font-size:1em',textInput("text53", label="Decimal:",".",width="100%"))
############################### A QUI. ######################################################################
),# column END
column(style='background: white',3,verticalLayout(tags$div(style= 'font-size:1em; font-weight: bold',"Dataset header"),
tags$div(style= 'font-size:1em;overflow: auto;height:25vh',tableOutput("contents51")))
),# column END
column(style='background: white',3,verticalLayout(tags$div(style= 'font-size:1em; font-weight: bold',"Dataset tail"),
tags$div(style= 'font-size:1em;overflow: auto;height:25vh',tableOutput("contents52")))
),# column END
),# fluidRow END
fluidRow(style='height:60vh;background: white;',
column(3,style='background: aliceblue',verticalLayout(actionButton(inputId = "go52",label = "Projection"),
tags$h4("Summary"),
tags$div(style= 'font-size:1em;',verbatimTextOutput("texto53",placeholder=TRUE)),
downloadButton("down51","Download Projection",class="butt"),
br(),
checkboxInput(inputId = "check52",label = "Load projection?",value=FALSE),
conditionalPanel(condition = "input.check52== 1",
tags$div(style= 'font-size:1em',fileInput("OtherOutput", label="Load Projection file:", accept=c("RData"),buttonLabel=div(style= CSS24,"Browse..."),width="90%"))),
conditionalPanel(condition = "input.check52== 1",
tags$div(style=CSS33,actionButton(inputId = "go53",label = div(style=CSS331,"Load")))),
conditionalPanel(condition = "input.check52== 1",br(),
tags$div(style= 'font-size:1em;',verbatimTextOutput("texto54",placeholder=TRUE)))
)),
column(2,verticalLayout(tags$h4("SOM (projection)"),
tags$div(withSpinner(plotOutput("plot51"),type=6,color="green"))
)),
column(2,verticalLayout(tags$h5("__________________"),
tags$div(style= 'font-size:1em;',selectInput("menu51", label="Feature", c(" ","hits","qerrs"), selected = " ",width="80%")),
uiOutput("UIinput51"),
tags$div(style= 'font-size:1em; font-weight: bold',numericInput("h51", "Height(in):", 3,width="80%")),
tags$div(style= 'font-size:1em; font-weight: bold',numericInput("w51", "Width(in):", 2,width="80%")),
downloadButton("down52","download pdf",class="butt"),
tags$h5(""),
downloadButton("down53","download Hits bs",class="butt"),
tags$h5(""),
downloadButton("down54","download Qerrs bs",class="butt")
)),
column(5,verticalLayout(fluidRow(column(5,tags$h4("Clusters (projection)")),
column(5,tags$div(style= 'font-size:1em',selectInput("menu53", label=NULL, c("Proj by cluster","Proj by variable"), selected = "Proj by cluster",width="90%")))),
withLoader(plotOutput("plot52"),type="text",loader=list(marquee("Calculating...",
direction="down", behavior = "scroll",width = "90%",style="font-size:15px; font-weight: bold; color:gray")))),
downloadButton("down55","download pdf",class="butt"))
)# fluidRow END
), # tabPanel END
#---------------------------------------------------------------------------------------------------------------------------------------------------------------------
# SIXTH TAB
#---------------------------------------------------------------------------------------------------------------------------------------------------------------------
tabPanel("Daily profiles", fluid = TRUE,
tags$style(CSS31), # Non toccare!!! Se viene tolto salta formattazione in tab 3 e altre parti (mah...)
fluidRow(style='height:25vh;background: aliceblue;',
# Input: File input and help ----
column(3,tags$div(style= 'font-size:1em;',selectInput("menu61", label="Dataset", c("training","projection"), selected = "training",width="80%")),
uiOutput("UIinput63")
),# column END
column(3,
uiOutput("UIinput61")
),# column END
column(4,
uiOutput("UIinput62"),
verbatimTextOutput("texto61", placeholder = TRUE)
############################### CAMBIATO DA QUI: ############################################################
),# column END
column(2,
tags$h2(""),
downloadButton("down66","Units cluster assignment",class="butt"),
tags$h4(""),
downloadButton("down67","Data cluster assignment",class="butt")
)# column END
############################### A QUI ############################################################
),# fluidRow END
fluidRow(style='height:60vh;background: white;',
column(6,verticalLayout(splitLayout(cellWidths = c("66%","3%","14%","17%"),
tags$h4("Daily graph"),
tags$h4(""),
tags$div(style= 'font-size:1em; font-weight: bold',"Xlab dim:"),
tags$div(style= CSS32,numericInput("x61", label=NULL, 0.75,width="100%"))),
tags$div(withSpinner(plotOutput("plot61"),type=6,color="green")),
splitLayout(cellWidths = c("7%","12%","7%","12%","9%","11%","3%","19%","20%"),
tags$div(style= 'font-size:1em; font-weight: bold',"H(in):"),
tags$div(style= CSS32,numericInput("h61", label=NULL, 5,width="100%")),
tags$div(style= 'font-size:1em; font-weight: bold',"W(in):"),
tags$div(style= CSS32,numericInput("w61", label=NULL, 7,width="100%")),
downloadButton("down61","pdf",class="butt"),
downloadButton("down62","table",class="butt"),
tags$h4(""),
downloadButton("down63","overall tab",class="butt"),
downloadButton("down64","monthly tab",class="butt")
))
),# column END
column(6,verticalLayout(
tags$h4("Sample profile"),
plotOutput("plot62")),
downloadButton("down65","pdf",class="butt")
)# column END
)# fluidRow END
) # tabPanel END
#-----------------------------------------
) # tabsetPanel END
) # mainPanel END
) # pageWithSidebar END
##########################################################################################################################################################
#
# INPUT & OUTPUT
#
##########################################################################################################################################################
# Define server logic to read selected file ----
server <- function(input, output,session) {
#---------------------------------------------------------------------------------------------------------------------------------------------------------------------
# server: FIRST TAB
#---------------------------------------------------------------------------------------------------------------------------------------------------------------------
#------Da qui richiamo input:
############################### CAMBIATO DA QUI: ############################################################
Experimental <- eventReactive(input$go11,{openair::import(input$Experimental$datapath,date="date",date.format=input$text11,
file.type = "txt", header = 1, sep = input$text12, dec=input$text13)})
############################### A QUI. ######################################################################
#--- Elimination of NA values:
ExpClean<-reactive({na.omit(data.frame(Experimental()))})
#------Da qui output:
output$contents11 <- renderTable({head(data.frame(date=as.character(ExpClean()[,1]),ExpClean()[,-1]))},spacing="xs")
output$contents12 <- renderTable({tail(data.frame(date=as.character(ExpClean()[,1]),ExpClean()[,-1]))},spacing="xs")
output$texto11<-renderText({paste("Number of uploaded rows =",as.character(nrow(ExpClean())))})
output$texto12<-renderText({paste("Number of deleted NA rows =",as.character(nrow(Experimental())-nrow(ExpClean())))})
CheckNum<-reactive({sum(sapply(ExpClean()[,c(2:ncol(ExpClean()))], is.numeric))==length(c(2:ncol(ExpClean())))})
output$texto13<-renderText({if (CheckNum()==FALSE){
paste0("WARNING: one or more variables are non-numeric, check your data!")}
})
#---------------------------------------------------------------------------------------------------------------------------------------------------------------------
# server: SECOND TAB
#---------------------------------------------------------------------------------------------------------------------------------------------------------------------
#--- Centering and scaling of experimental data:
prepare<-reactive({scale(ExpClean()[,-1],center = TRUE,scale=TRUE)})
CenterVal<-reactive({attr(prepare(),"scaled:center")})
ScaleVal<-reactive({attr(prepare(),"scaled:scale")})
dataset<-reactive({prepare()}) # matrice da usare per il calcolo della SOM
#--- Default SOM map dimensions:
DimsD<-reactive({SOMEnv::som_dimR(dataset(),type=input$Size)}) #QUESTO dipende da un input "Size", quindi non serve "Restore Button", basta ricambiare "Size" e si torna ai default!!!!
output$UIinput21 <- renderUI({if(input$go11==0) {
tagList(
column(3,div(style= 'font-size:1em',textInput("txtfake21", "Number of SOM map rows:", value="Waiting for file input...",width="90%")),
div(style= 'font-size:1em',textInput("txtfake22", "Number of SOM map cols:", value="Waiting for file input...",width="90%"))))
} else {
tagList(
column(3,div(style= 'font-size:1em',numericInput("Row", "Number of SOM map rows:", value=DimsD()$Row,min=2,width="90%")),
div(style= 'font-size:1em',numericInput("Col", "Number of SOM map cols:", value=DimsD()$Col,min=2,width="90%"))))}
})
#--- SOM map dimensions evaluated from inputs:
Dims<-reactive({list(Row=input$Row,Col=input$Col,munits=input$Row*input$Col)})
output$texto21<-renderText({paste("Number of units =",as.character(Dims()$munits))})
#--- Parameters evaluated from inputs:
Init<-reactive({SOMEnv::som_initR(dataset(),Dims()$Row,Dims()$Col,Dims()$munits)})
dlen<-reactive({nrow(dataset())})
mpd<-reactive({Dims()$munits/dlen()})
#--- Default parameters evaluated from inputs:
EpochsD<-reactive({max(2,(ceiling(mpd()*10)+ceiling(mpd()*40)))}) # da rough+finetune phase by Vesanto
radMaxD<-reactive({max(1,ceiling(max(c(Dims()$Row,Dims()$Col,na.rm=T))/8))}) # da finetune phase by Vesanto
RadDtext<-reactive({paste(radMaxD(),",",1,sep="")})
output$UIinput22 <- renderUI({tagList(
column(2,div(style= 'font-size:1em',numericInput("Epochs", "Number of epochs:", value=EpochsD(),min=2,width="90%")))
)})
output$UIinput23 <- renderUI({tagList(
div(style= 'font-size:1em',textInput("Rad", "Neigh.radius:", RadDtext(),width="90%"))
)})
#--- Parameters rearranged from inputs:
Rad<-reactive({as.numeric(unlist(strsplit(input$Rad, ',')))})
###------------- SOM MAKE
#------- SOM grid
som_grid <- eventReactive(input$go21,{somgrid(xdim = input$Row, ydim=input$Col, topo="hexagonal", toroidal = FALSE,neighbourhood.fct = input$Neigh)})
#--- SOM training:
som_model <- eventReactive(input$go21,{som(dataset(),
grid=som_grid(),
radius=Rad(),
rlen=input$Epochs,
init=Init(),
dist.fcts="euclidean", #fixed
mode="pbatch",#fixed
keep.data = F)})
#--- Training data projection:
Training<-reactive({map(som_model(), dataset())})
Bmus<-reactive({Training()$unit.classif})
Qerrs<-reactive({Training()$distances})
############################### CAMBIATO DA QUI (20201008): ############################################################
#--- Hits
BmusFreq<-reactive({factor(Bmus(), levels=c(1:Dims()$munits))})
TabHits<-reactive({data.frame(table(BmusFreq()))})
Hits<-reactive({TabHits()$Freq})
############################### A QUI (20201008) #######################################################################
#--- Denormalized codebook
codebook<-reactive({data.frame(som_model()$codes)})
DeCod <- reactive({data.frame(t(apply(codebook(), 1, function(r)r*ScaleVal()+ CenterVal())))})
#--- Prototype x and y coordinates
Coord<-eventReactive(input$go21,{SOMEnv::CodeCoord(input$Row,input$Col)})
#--- Umatrix
Umat<-reactive({SOMEnv::som_umatR(codebook(), input$Row,input$Col)})
#--- Output:
############################### CAMBIATO DA QUI: ############################################################
output$contents21 <- renderTable({head(data.frame(Unit=c(1:nrow(DeCod())),DeCod()))},spacing="xs")
############################### A QUI ######################################################################
output$texto22<-renderPrint({
cat("SOM of size ", Dims()$Row, "x", Dims()$Col, sep = "")
cat("\n Number of samples used for training: ",nrow(ExpClean()),sep = "")
cat("\n Number of prototypes(units): ",nrow(codebook()),sep = "")
cat("\n Topology: hexagonal",sep = "")
cat("\n Neighbourhood function: ",input$Neigh, sep = "")
cat("\n Distance measure used: Euclidean",sep = "")
cat("\n Learning algorithm: pbatch",sep = "")
cat("\n Number of training epochs: ",input$Epochs,sep = "")
})
#--- Save list:
OutputRdata<-reactive({list(ExpClean=ExpClean(),dataset=dataset(),ScaleVal=ScaleVal(),CenterVal=CenterVal(),Epochs=input$Epochs,som_model=som_model(),
Neigh=input$Neigh,Dims=Dims(),Coord=Coord(),codebook=codebook(),DeCod=DeCod(),Umat=Umat(),Hits=Hits(),Bmus=Bmus(),Qerrs=Qerrs())})
output$down21<-downloadHandler(filename=function() {paste0("SOM_output_",format(Sys.time(),format="%Y-%m-%d_%H-%M"),".RData")},
content = function(file) {
SOMoutput<-OutputRdata()
list.save(SOMoutput,file=file)}
)
############################### AGGIUNTO DA QUI: ############################################################
#-- Map quality evaluation
QE<-eventReactive(input$go22,{round(sum(Qerrs())/nrow(ExpClean()),digit=6)})
output$texto23<-renderPrint({cat(QE())})
############################### CAMBIATO DA QUI (20201008): ############################################################
TE<-eventReactive(input$go23,{SOMEnv::SOMtopol(as.matrix(dataset()),as.matrix(codebook()),as.matrix(som_model()$grid$pts))})
output$texto24<-renderPrint({cat(TE())})
############################### A QUI (20201008) #######################################################################
#---------------------------------------------------------------------------------------------------------------------------------------------------------------------
# server: THIRD TAB
#---------------------------------------------------------------------------------------------------------------------------------------------------------------------
SOMoutput <- eventReactive(input$go31,{list.load(input$SOMoutput$datapath)})
output$texto31<-renderPrint({
cat("SOM of size ", SOMoutput()$Dims$Row, "x", SOMoutput()$Dims$Col, sep = "")
cat("\n Number of samples used for training: ",nrow(SOMoutput()$ExpClean),sep = "")
cat("\n Number of variables: ",ncol(SOMoutput()$codebook),sep = "")
cat("\n Neighbourhood function: ",SOMoutput()$Neigh, sep = "")
cat("\n Number of training epochs: ",SOMoutput()$Epochs,sep = "")
})
#-- Heatmaps
MxRow<-eventReactive(input$go31,{if (ncol(SOMoutput()$codebook)<=14) {2} else {3} })
MxCol<-eventReactive(input$go31,{
if(ncol(SOMoutput()$codebook)%%MxRow()==0) {ncol(SOMoutput()$codebook)%/%MxRow()
} else {ncol(SOMoutput()$codebook)%/%MxRow()+1}
})
output$UIinput31 <- renderUI({tagList(splitLayout(style= 'overflow-x: hidden',cellWidths = c("25%","25%","25%","25%"),
div(style= 'font-size:1em; font-weight: bold',"Map x row:"),
div(style= CSS32,numericInput("n31", label=NULL, MxRow(),width="80%")),
div(style= 'font-size:1em; font-weight: bold',"Map x col:"),
div(style= CSS32,numericInput("n32", label=NULL, MxCol(),width="80%"))
)
)})
Table30<-function() {SumTable<-SOMEnv::paramQuant(SOMoutput()$DeCod[,1]); ##################################----- de-normalized!
for (i in c(2:ncol(SOMoutput()$DeCod))) {
P<-SOMEnv::paramQuant(SOMoutput()$DeCod[,i]);SumTable<-cbind(SumTable,P[,2])}
colnames(SumTable)<-c("Statistic",colnames(SOMoutput()$DeCod))
SumTable<-data.frame(SumTable)
return(SumTable)
}
output$down30 <- downloadHandler(filename ="Codebook_basic statistics.txt",
content = function(file) {
write.table(Table30(), file, row.names = F,col.names=T,sep="\t",quote=F)}
)
Plot31<- function() {if (is.null(input$SOMoutput)) {return(NULL)
} else {SOMEnv::HexagonsVar(c(input$n31,input$n32),SOMoutput()$codebook,SOMoutput()$Coord,SOMoutput()$Dims$Row,SOMoutput()$Dims$Col)}}
output$plot31<- renderPlot({if (input$go32[[1]] == 0) return(); Plot31()})
output$down31 <- downloadHandler(filename ="Heatmaps.pdf",
content = function(file) {
pdf(file,height=input$h31,width=input$w31)
Plot31()
dev.off()}
)
#-- SOM map features
output$UIinput32 <- renderUI({if(input$menu31==" ") {
tagList(div(style= 'font-size:1em;',selectInput("menu32", label="Plot type", c(" "), selected = " ",width="80%")))
} else if (input$menu31=="umat"){tagList(div(style= 'font-size:1em;',selectInput("menu32", label="Plot type", c("bw","gs"), selected = "gs",width="80%")))
} else if (input$menu31=="hits" | input$menu31=="qerrs"){tagList(div(style= 'font-size:1em;',
selectInput("menu32", label="Plot type", c("grayscale","black filling"), selected = "grayscale",width="80%")))}
})
Switch3<-reactive({paste0(input$menu31,input$menu32)})
Plot32<- function() {switch(Switch3(),
" "={return(NULL)},
"umatgs"={SOMEnv::UmatGraph(SOMoutput()$Umat,SOMoutput()$Dims$Row,SOMoutput()$Dims$Col,colorscale="gs")},
"umatbw"={SOMEnv::UmatGraph(SOMoutput()$Umat,SOMoutput()$Dims$Row,SOMoutput()$Dims$Col,colorscale="bw")},
"hitsgrayscale"={SOMEnv::HexaHitsQuant(SOMoutput()$Hits,SOMoutput()$Coord,SOMoutput()$Dims$Row,SOMoutput()$Dims$Col)},
"hitsblack filling"={SOMEnv::HexaHits(SOMoutput()$Hits,SOMoutput()$Coord,SOMoutput()$Dims$Row,SOMoutput()$Dims$Col)},
"qerrsgrayscale"={SOMEnv::HexaQerrsQuant(SOMoutput()$Bmus,SOMoutput()$Qerrs,SOMoutput()$Coord,SOMoutput()$Dims$Row,SOMoutput()$Dims$Col)},
"qerrsblack filling"={SOMEnv::HexaQerrs(SOMoutput()$Bmus,SOMoutput()$Qerrs,SOMoutput()$Coord,SOMoutput()$Dims$Row,SOMoutput()$Dims$Col)})
}
output$plot32<- renderPlot({Plot32()})
output$down32 <- downloadHandler(filename =function() {paste0("SOM_",input$menu31,"_",input$menu32,".pdf")},
content = function(file) {
pdf(file,height=input$h32,width=input$w32)
Plot32()
dev.off()}
)
#-- Basic statistics
Table31<-function() {SOMEnv::paramQuant(SOMoutput()$Hits)}
output$down33 <- downloadHandler(filename ="Hits_basic statistics.txt",
content = function(file) {
write.table(Table31(), file, row.names = F,col.names=T,sep="\t",quote=F)}
)
Table32<-function() {SOMEnv::paramQuant(SOMoutput()$Qerrs)}
output$down34 <- downloadHandler(filename ="Qerrs_basic statistics.txt",
content = function(file) {
write.table(Table32(), file, row.names = F,col.names=T,sep="\t",quote=F)}
)
#---------------------------------------------------------------------------------------------------------------------------------------------------------------------
# server: FOURTH TAB
#---------------------------------------------------------------------------------------------------------------------------------------------------------------------
# su dati caricati in Third Tab!!!!
Switch41<-function() {switch(input$choice41,
"Yes"={paste0("Yes")},
"No"={paste0("No")})
}
Kmeans <- eventReactive(input$go41,{if (Switch41()=="No"){SOMEnv::kmeans_clustersRProg(SOMoutput()$codebook,k=input$n41,times=input$n42)
} else {SOMEnv::kmeans_clustersRProg(SOMoutput()$codebook,k=input$n41,times=input$n42,seed=input$Seed) }
})
#-- Risultati aggiuntivi:
############################### CORRETTO DA QUI: 20210211 ###############################################
BCentr<-reactive({BmusCentr(Kmeans()$centroids,SOMoutput()$som_model,length(Kmeans()$centroids))})
ClusExp<-function() {D<-rep(0,nrow(SOMoutput()$ExpClean))
for (i in c(2:length(Kmeans()$centroids))) {d<-BmusClus(SOMoutput()$Bmus,Kmeans()$clusNum[,i]);
D<-cbind(D,d)}; return(data.frame(D))}
############################### A QUI. ######################################################################
#-- Output:
OutputKdata<-reactive({list(centroids=Kmeans()$centroids,clusNum=data.frame(Kmeans()$clusNum),ind=Kmeans()$ind,
err=Kmeans()$err,seed=Kmeans()$seed,BCentr=BCentr(),ClusExp=ClusExp())})
output$down41<-downloadHandler(filename=function() {paste0("Kmeans_output_",format(Sys.time(),format="%Y-%m-%d_%H-%M"),".RData")},
content = function(file) {
KmeansOutput<-OutputKdata()
list.save(KmeansOutput,file=file)}
)
#-- If KmeansOutput file is loaded:
Kmeansoutput <- eventReactive(input$go42,{list.load(input$KmeansOutput$datapath)})
KmeansForPlot<-reactive({if(input$check41==FALSE){OutputKdata()} else {Kmeansoutput()} })
#-- Aggiornamento input in base a cluster migliore:
BestClus<-reactive({if(is.null(KmeansForPlot())) {5} else {which(KmeansForPlot()$ind==min(KmeansForPlot()$ind,na.rm=T))+1}})
output$UIinput42 <- renderUI({tagList(
fluidRow(
column(2,div(style= 'font-size:1em',numericInput("n43", label="N.clusters:", BestClus(),min=2,width="80%"))),
column(1,colourpicker::colourInput("col40", label=div(style= CSS34,"Cl 1"), "red",showColour="background",returnName=T,palette="limited")),
column(1,colourpicker::colourInput("col41", label=div(style= CSS34,"Cl 2"), "yellow",showColour="background",returnName=T,palette="limited")),
conditionalPanel("input.n43>=3",column(1,colourpicker::colourInput("col42", label=div(style= CSS34,"Cl 3"), "green",showColour="background",returnName=T,palette="limited"))),
conditionalPanel("input.n43>=4",column(1,colourpicker::colourInput("col43", label=div(style= CSS34,"Cl 4"), "cyan",showColour="background",returnName=T,palette="limited"))),
conditionalPanel("input.n43>=5",column(1,colourpicker::colourInput("col44", label=div(style= CSS34,"Cl 5"), "darkorange1",showColour="background",returnName=T,palette="limited"))),
conditionalPanel("input.n43>=6",column(1,colourpicker::colourInput("col45", label=div(style= CSS34,"Cl 6"), "magenta",showColour="background",returnName=T,palette="limited"))),
conditionalPanel("input.n43>=7",column(1,colourpicker::colourInput("col46", label=div(style= CSS34,"Cl 7"), "aquamarine",showColour="background",returnName=T,palette="limited"))),
conditionalPanel("input.n43>=8",column(1,colourpicker::colourInput("col47", label=div(style= CSS34,"Cl 8"), "sandybrown",showColour="background",returnName=T,palette="limited"))),
column(2,div(style=CSS33,actionButton(inputId = "go43",label = div(style=CSS331,"Plot"))))
) # fluidRow END
)})
TextClus<-reactive({if(is.null(KmeansForPlot())) {paste(c(1:5),collapse=",",sep="")} else {paste(c(1:input$n43),collapse=",",sep="")}})
output$UIinput41 <- renderUI({tagList(
div(style= CSS32,textInput(inputId = "text41",label = "Cluster numbers:",value=TextClus(),width="90%"))
)})
#-- Output DB-index plot:
Plot41<-function(){if (is.null(KmeansForPlot())) {return(NULL)
} else {
opar <- par(mar=c(4,3,1,1),oma=c(1,1,1,1))
plot(c(2:(length(KmeansForPlot()$ind)+1)),KmeansForPlot()$ind,col="gray",xlim=c(1,length(KmeansForPlot()$ind)+2),type="h",
ylim=c(min(KmeansForPlot()$ind,na.rm=T)-0.05*min(KmeansForPlot()$ind,na.rm=T),max(KmeansForPlot()$ind,na.rm=T)+0.05*max(KmeansForPlot()$ind,na.rm=T)),
xlab="Cluster number",ylab="DB-index",xaxt="n")
text(c(2:(length(KmeansForPlot()$ind)+1)),KmeansForPlot()$ind,round(KmeansForPlot()$ind,digit=2),cex=0.75,pos=3)
axis(1,at=seq(2,length(KmeansForPlot()$ind)+1,1),labels=NA,tcl=-0.3,cex.axis=0.6)
axis(1,at=seq(2,length(KmeansForPlot()$ind)+1,1),labels=seq(2,length(KmeansForPlot()$ind)+1,1),lwd=0,line=-0.2,cex.axis=1)}
on.exit(par(opar))
}
output$plot41<-renderPlot({if (is.null(KmeansForPlot())) {return(NULL)
} else {Plot41()}})
output$down42 <- downloadHandler(filename =function() {paste0("DB-index_plot",".pdf")},
content = function(file) {
pdf(file,height=input$h41,width=input$w41)
Plot41()
dev.off()}
)
#-- Output SOM cluster map plot:
colSeq4<-reactive({rep_len(c(input$col40,input$col41,input$col42,input$col43,input$col44,input$col45,
input$col46,input$col47), input$n43)})
numSeq41<-reactive({as.numeric(unlist(strsplit(input$text41, ',')))})
numSeq42<-reactive({as.numeric(unlist(strsplit(input$text42, ',')))})
numSeq43<-reactive({c(input$y41,input$y42)})
Centroids4<-reactive({data.frame(KmeansForPlot()$centroids[[input$n43]][order(numSeq41()),])})
############################### CORRETTO DA QUI: 20210211 ###############################################
clusNum4<-reactive({SOMEnv::NClusChange(KmeansForPlot()$clusNum[,input$n43],Centroids4(),numSeq41())})
BCentr4<-reactive({KmeansForPlot()$BCentr[[input$n43]][order(numSeq41())]})
ClusExp4<-reactive({SOMEnv::NClusChange(KmeansForPlot()$ClusExp[,input$n43],Centroids4(),numSeq41())})
############################### A QUI. ######################################################################
Plot42<- function() {if (is.null(KmeansForPlot())) {return(NULL)
} else if (is.null(SOMoutput())) {return(NULL)
} else if (length(numSeq41())!=input$n43) {return(NULL)
} else {
opar <- par(mar=c(1,1,1,2))
SOMEnv::HexagonsClus(Centroids4(), clusNum4(), BCentr4(),
SOMoutput()$Coord, SOMoutput()$Dims$Row, SOMoutput()$Dims$Col, colSeq = colSeq4())}
on.exit(par(opar))
}
output$plot42<-renderPlot({if (is.null(KmeansForPlot())|input$go43==0 ) {return(NULL)
} else if (is.null(SOMoutput())) {return(NULL)
} else {Plot42()}
})
output$down43 <- downloadHandler(filename =function() {paste0("SOM map",".pdf")},
content = function(file) {
pdf(file,height=input$h41,width=input$w41)
Plot42()
dev.off()}
)
#-- Output cluster profiles plot:
output$UIinput43 <- renderUI({if(input$menu41=="by cluster"| input$menu41=="Exp by cluster") {
tagList(
column(7,splitLayout(cellWidths = c("8%","7%","10%","7%","10%","7%","12%","10%","12%","20%"),
h5(""),
div(style= 'font-size:1em; font-weight: bold',"Y min:"),
############################### CORRETTO DA QUI: 20210211 ###############################################
div(style= CSS32,numericInput("y41", label=NULL, -2,width="100%")),
div(style= 'font-size:1em; font-weight: bold',"Y max:"),
div(style= CSS32,numericInput("y42", label=NULL, 2,width="100%")),
############################### A QUI. ######################################################################
div(style= 'font-size:1em; font-weight: bold',"Y grid:"),
div(style= CSS32,textInput(inputId = "text42",label=NULL,value="-1,0,1",width="100%")),
div(style= 'font-size:1em; font-weight: bold',"Xlab dim:"),
div(style= CSS32,numericInput("x41", label=NULL, 0.75,width="100%")),
downloadButton("down44","pdf",class="butt")))
)
} else if(input$menu41=="by variable"| input$menu41=="Exp by variable"){
tagList(
column(7,splitLayout(cellWidths = c("15%","15%","15%","15%","15%","25%"),
h5(""),
div(style= 'font-size:1em; font-weight: bold',"Map x row:"),
div(style= CSS32,numericInput("n44", label=NULL, MxRow(),width="80%")),
div(style= 'font-size:1em; font-weight: bold',"Map x col:"),
div(style= CSS32,numericInput("n45", label=NULL, MxCol(),width="80%")),
downloadButton("down44","pdf",class="butt")))
)}
})
############################### CORRETTO DA QUI: 20210211 ###############################################
Plot43<- function() {if (is.null(KmeansForPlot())) {return(NULL)
} else if (is.null(SOMoutput())) {return(NULL)
} else if (length(numSeq41())!=input$n43) {return(NULL)
} else {switch(input$menu41,
"by cluster"={SOMEnv::BoxUnits(SOMoutput()$codebook,clusNum4(),Centroids4(),Ylim=numSeq43(),pitch=numSeq42(),xdim=input$x41)},
"by variable"={SOMEnv::BoxClus(c(input$n44,input$n45),SOMoutput()$DeCod,clusNum4(),Centroids4())},
"Exp by cluster"={SOMEnv::BoxUnits(SOMoutput()$dataset,ClusExp4(),Centroids4(),Ylim=numSeq43(),pitch=numSeq42(),xdim=input$x41)},
"Exp by variable"={SOMEnv::BoxClus(c(input$n44,input$n45),SOMoutput()$ExpClean[,-1],ClusExp4(),Centroids4())})}
}
############################### A QUI. ######################################################################
output$plot43<-renderPlot({if (is.null(KmeansForPlot())|input$go43==0 ) {return(NULL)
} else if (is.null(SOMoutput())) {return(NULL)
} else {Plot43()}
})
output$down44 <- downloadHandler(filename =function() {paste0("Cluster profiles",".pdf")},
content = function(file) {
pdf(file,height=input$h41,width=input$w41)
Plot43()
dev.off()}
)
#---------------------------------------------------------------------------------------------------------------------------------------------------------------------
# server: FIFTH TAB
#---------------------------------------------------------------------------------------------------------------------------------------------------------------------
#------Da qui richiamo input Other:
############################### CAMBIATO DA QUI: ############################################################
Other <- eventReactive(input$go51,{openair::import(input$Other$datapath,date="date",date.format=input$text51,
file.type = "txt",header = 1,sep = input$text52,dec=input$text53)})
############################### A QUI #######################################################################
#--- Elimination of NA values:
OtherClean<-reactive({na.omit(data.frame(Other()))})
#------Da qui output:
output$contents51 <- renderTable({head(data.frame(date=as.character(OtherClean()[,1]),OtherClean()[,-1]))},spacing="xs")
output$contents52 <- renderTable({tail(data.frame(date=as.character(OtherClean()[,1]),OtherClean()[,-1]))},spacing="xs")
output$texto51<-renderText({paste("Number of uploaded rows =",as.character(nrow(OtherClean())))})
output$texto52<-renderText({paste("Number of deleted NA rows =",as.character(nrow(Other())-nrow(OtherClean())))})
#--- Data scaling & projection:
newdata<-eventReactive(input$go52,{as.matrix(scale(OtherClean()[,-1],center = SOMoutput()$CenterVal, scale = SOMoutput()$ScaleVal))})
Projection<-eventReactive(input$go52,{if (ncol(OtherClean())!=ncol(SOMoutput()$ExpClean)) {return(NULL)} else {map(SOMoutput()$som_model, newdata())}})
#--- Projection Outputs
BmusPROJ<-reactive({Projection()$unit.classif})
QerrsPROJ<-reactive({Projection()$distances})
############################### CORRETTO DA QUI: 20210211 ###############################################
BmusFreqPROJ<-reactive({factor(BmusPROJ(), levels=c(1:SOMoutput()$Dims$munits))})
TabHitsPROJ<-reactive({data.frame(table(BmusFreqPROJ()))})
HitsPROJ<-reactive({TabHitsPROJ()$Freq})
############################### A QUI. ######################################################################
output$texto53<-renderPrint({if (input$go52==0) {cat("", sep = "")
} else if (ncol(OtherClean())!=ncol(SOMoutput()$ExpClean)) {cat("Upload a file with ", ncol(SOMoutput()$ExpClean), " columns!", sep = "")
} else { cat("Projection of ", length(BmusPROJ()), " samples on", sep = "")
cat("\n SOM of size ", SOMoutput()$Dims$Row, "x", SOMoutput()$Dims$Col, sep = "")
cat("\n Topology: hexagonal",sep = "")
cat("\n Neighbourhood function: ",SOMoutput()$Neigh, sep = "")
cat("\n Distance measure used: Euclidean",sep = "")
cat("\n Number of training epochs: ",SOMoutput()$Epochs,sep = "")}
})
#---- Cluster assignment (respect to Kmeans graphs in third tab!)
############################### CORRETTO DA QUI: 20210211 ###############################################
ClusProj<-function() {D<-rep(0,length(BmusPROJ()))
for (i in c(2:length(KmeansForPlot()$centroids))) {d<-BmusClus(BmusPROJ(),KmeansForPlot()$clusNum[,i]);
D<-cbind(D,d)}; return(data.frame(D))}
############################### A QUI. ######################################################################
#--- Save Projection Outputs
OutputPdata<-reactive({list(OtherClean=OtherClean(),newdata=newdata(),HitsPROJ=HitsPROJ(),BmusPROJ=BmusPROJ(),QerrsPROJ=QerrsPROJ(),ClusProj=ClusProj())})
output$down51<-downloadHandler(filename=function() {paste0("Projection_output_",format(Sys.time(),format="%Y-%m-%d_%H-%M"),".RData")},
content = function(file) {
SOMoutput<-OutputPdata()
list.save(SOMoutput,file=file)}
)
#-- If ProjectionOutput file is loaded:
ProjectionOutput <- eventReactive(input$go53,{list.load(input$OtherOutput$datapath)})
ProjForPlot<-reactive({if(input$check52==FALSE){OutputPdata()} else {ProjectionOutput()} })
output$texto54<-renderPrint({if(input$go53==0){cat("", sep = "")} else {cat("Projection of ", length(ProjForPlot()$BmusPROJ), " samples loaded", sep = "")}
})
#-- SOM map features (Projection)
output$UIinput51 <- renderUI({if(input$menu51==" ") {
tagList(div(style= 'font-size:1em;',selectInput("menu52", label="Plot type", c(" "), selected = " ",width="80%")))
} else if (input$menu51=="hits" | input$menu51=="qerrs"){tagList(div(style= 'font-size:1em;',
selectInput("menu52", label="Plot type", c("grayscale","black filling"), selected = "grayscale",width="80%")))}
})
Switch5<-reactive({paste0(input$menu51,input$menu52)})
Plot51<- function() {switch(Switch5(),
" "={return(NULL)},
"hitsgrayscale"={SOMEnv::HexaHitsQuant(ProjForPlot()$HitsPROJ,SOMoutput()$Coord,SOMoutput()$Dims$Row,SOMoutput()$Dims$Col)},
"hitsblack filling"={SOMEnv::HexaHits(ProjForPlot()$HitsPROJ,SOMoutput()$Coord,SOMoutput()$Dims$Row,SOMoutput()$Dims$Col)},
"qerrsgrayscale"={SOMEnv::HexaQerrsQuant(ProjForPlot()$BmusPROJ,ProjForPlot()$QerrsPROJ,SOMoutput()$Coord,SOMoutput()$Dims$Row,SOMoutput()$Dims$Col)},
"qerrsblack filling"={SOMEnv::HexaQerrs(ProjForPlot()$BmusPROJ,ProjForPlot()$QerrsPROJ,SOMoutput()$Coord,SOMoutput()$Dims$Row,SOMoutput()$Dims$Col)})
}
output$plot51<- renderPlot({Plot51()})
output$down52 <- downloadHandler(filename =function() {paste0("SOM_projection_",input$menu51,"_",input$menu52,".pdf")},
content = function(file) {
pdf(file,height=input$h51,width=input$w51)
Plot51()
dev.off()}
)
#-- Basic statistics
Table51<-function() {SOMEnv::paramQuant(ProjForPlot()$HitsPROJ)}
output$down53 <- downloadHandler(filename ="Hits_projection_basic statistics.txt",
content = function(file) {
write.table(Table51(), file, row.names = F,col.names=T,sep="\t",quote=F)}
)
Table52<-function() {SOMEnv::paramQuant(ProjForPlot()$QerrsPROJ)}
output$down54 <- downloadHandler(filename ="Qerrs_projection_basic statistics.txt",
content = function(file) {
write.table(Table52(), file, row.names = F,col.names=T,sep="\t",quote=F)}
)
#-- Cluster profiles (projection) --> respect to fourth tab parameters!!!
############################### CORRETTO DA QUI: 20210211 ###############################################
ClusProj4<-reactive({SOMEnv::NClusChange(ProjForPlot()$ClusProj[,input$n43],Centroids4(),numSeq41())})
Plot52<- function() {if (is.null(ProjForPlot())) {return(NULL)
} else if (is.null(ProjForPlot())) {return(NULL)
} else if (length(numSeq41())!=input$n43) {return(NULL)
} else {switch(input$menu53,
"Proj by cluster"={SOMEnv::BoxUnits(ProjForPlot()$newdata,ClusProj4(),Centroids4(),Ylim=numSeq43(),pitch=numSeq42(),xdim=input$x41)},
"Proj by variable"={SOMEnv::BoxClus(c(input$n44,input$n45),ProjForPlot()$OtherClean[,-1],ClusProj4(),Centroids4())})}
}
############################### A QUI. ######################################################################
output$plot52<-renderPlot({if (is.null(ProjForPlot())) {return(NULL)
} else {Plot52()}
})
output$down55 <- downloadHandler(filename =function() {paste0("Cluster profiles_projection",".pdf")},
content = function(file) {
pdf(file,height=input$h51,width=input$w51)
Plot52()
dev.off()}
)
#---------------------------------------------------------------------------------------------------------------------------------------------------------------------
# server: SIXTH TAB
#---------------------------------------------------------------------------------------------------------------------------------------------------------------------
#--- Scelta dataset
ForDailyGraph<- function() {switch(input$menu61,
"training"={data.frame(date=SOMoutput()$ExpClean$date,Cluster=ClusExp4(),Bmus=SOMoutput()$Bmus,Qerrs=SOMoutput()$Qerrs)},
"projection"={data.frame(date=ProjForPlot()$OtherClean$date,Cluster=ClusProj4(),Bmus=ProjForPlot()$BmusPROJ,Qerrs=ProjForPlot()$QerrsPROJ)})
}
#--- Guess Obs/Day
Guess<-reactive({format(ForDailyGraph()$date,"%d/%m/%Y")})
Guess2<-reactive({ordered(Guess(), levels = unique(Guess()))})
Count<-reactive({plyr::count(Guess2())})
Total<-reactive({round(median(Count()$freq,na.rm=T),digit=0)})
output$UIinput63 <- renderUI({tagList(div(style= CSS32,numericInput("n61", label="Obs/Day", Total(),width="80%"))
)})
#--- Scelta intervallo dati
dateStart61<-function() {paste(
substr(ForDailyGraph()[1,"date"], 9, 10),"/",
substr(ForDailyGraph()[1,"date"], 6, 7),"/",
substr(ForDailyGraph()[1,"date"], 1, 4),sep="")}
dateEnd61<-function() {paste(
substr(ForDailyGraph()[nrow(ForDailyGraph()),"date"], 9, 10),"/",
substr(ForDailyGraph()[nrow(ForDailyGraph()),"date"], 6, 7),"/",
substr(ForDailyGraph()[nrow(ForDailyGraph()),"date"], 1, 4),sep="")}
output$UIinput61 <- renderUI({tagList(
div(style= CSS32,textInput(inputId = "text61",label = "Starting date:",value=dateStart61(),width="80%")),
div(style= CSS32,textInput(inputId = "text62",label = "Ending date:",value=dateEnd61(),width="80%"))
)})
Selection<-reactive({selectByDate(ForDailyGraph(),start=input$text61,end=input$text62)})
#--- Daily Graph
############################### CORRETTO DA QUI: 20210211 ###############################################
Plot61<- function() {SOMEnv::DailyBar(Selection(),Selection()$Cluster,Centroids4(),colSeq=colSeq4(),Total=input$n61,xdim=input$x61,ydim=0.8)}
############################### A QUI. ######################################################################
output$plot61<- renderPlot({Plot61()})
output$down61 <- downloadHandler(filename =function() {paste0("Daily graph",".pdf")},
content = function(file) {
pdf(file,height=input$h61,width=input$w61)
Plot61()
dev.off()}
)
#--- Download tables:
############################### CORRETTO DA QUI: 20210211 ###############################################
Table61<-function() {SOMEnv::FreqD(Selection()$date,Selection()$Cluster,Centroids4(),Total=input$n61)}
############################### A QUI. ######################################################################
output$down62 <- downloadHandler(filename ="Interval selection.txt",
content = function(file) {
write.table(Table61(), file, row.names = T,col.names=T,sep="\t",quote=F)}
)
############################### CORRETTO DA QUI: 20210211 ###############################################
Table62<-function() {switch(input$menu61,
"training"={SOMEnv::Freq(ClusExp4(),Centroids4())},
"projection"={SOMEnv::Freq(ClusProj4(),Centroids4())})
}
############################### A QUI. ######################################################################
output$down63 <- downloadHandler(filename ="Overall.txt",
content = function(file) {
write.table(Table62(), file, row.names = F,col.names=T,sep="\t",quote=F)}
)
############################### CORRETTO DA QUI: 20210211 ###############################################
Table63<-function() {SOMEnv::FreqM(ForDailyGraph()$date,ForDailyGraph()$Cluster,Centroids4())}
############################### A QUI. ######################################################################
output$down64 <- downloadHandler(filename ="Monthly.txt",
content = function(file) {
write.table(Table63(), file, row.names = F,col.names=T,sep="\t",quote=F)}
)
############################### CORRETTO DA QUI: 20210211 ###############################################
Table64<-function() {data.frame(Unit=c(1:length(clusNum4())),Cluster=clusNum4())}
############################### A QUI. ######################################################################
output$down66 <- downloadHandler(filename ="Unit cluster assignment.txt",
content = function(file) {
write.table(Table64(), file, row.names = F,col.names=T,sep="\t",quote=F)}
)
Table65<-function() {data.frame(ForDailyGraph())}
output$down67 <- downloadHandler(filename =function() {paste0(as.character(input$menu61)," data cluster assignment.txt")},
content = function(file) {
write.table(Table65(), file, row.names = F,col.names=T,sep="\t",quote=F)}
)
#--- Single sample profile
output$UIinput62 <- renderUI({tagList(
div(style= CSS32,textInput(inputId = "text63",label = "Select sample:",value=as.character(ForDailyGraph()[2,"date"]),width="100%"))
)})
#--- Starting plot
SelDay<-reactive({ForDailyGraph()[which(as.character(ForDailyGraph()$date)==input$text63),]})
output$texto61<-renderPrint({ cat(" BMU = ", SelDay()[,"Bmus"],sep = "")
cat("\n Cluster = ", SelDay()[,"Cluster"], sep = "")
cat("\n Qe = ", round(SelDay()[,"Qerrs"],digit=3), sep = "")
})
SelExpClean<- function() {switch(input$menu61,
"training"={data.frame(date=SOMoutput()$ExpClean$date,SOMoutput()$dataset)},
"projection"={data.frame(date=ProjForPlot()$OtherClean$date,ProjForPlot()$newdata)})
}
SelBOX<-reactive({SelExpClean()[which(as.character(SelExpClean()$date)==input$text63),]})
SingleSampPlot<- function() {
SLN<-SelDay()[,3]
CLS<-SelDay()[,2]
CexLy<-0.9
CexAxy<-0.85
Ylab<-c("Cluster")
Ymin<-0.5
Ymax<-input$n43+0.5
nClus<-input$n43
Names<-colnames(SOMoutput()$codebook)
############################### CORRETTO DA QUI: 20210211 ###############################################
Xlim<-c(0.5,ncol(SOMoutput()$codebook)+0.5)
############################### A QUI. ######################################################################
Ylim<-numSeq43()
pitch<-numSeq42()
Xdim<-input$x41
i<-1
#---qua il primo grafico:
opar <- par(fig=c(0, 0.5, 0, 1),oma=c(1,1,1,1),mar=c(1,1.2,1,1.2),xpd=TRUE,pty="m",family="serif");
SOMEnv::HexagonsClus(Centroids4(), clusNum4(), BCentr4(),
SOMoutput()$Coord, SOMoutput()$Dims$Row, SOMoutput()$Dims$Col, colSeq = colSeq4())
mtext(SelDay()[,"date"][i],side=3,lin=0.5);
#--qua ultimo dato:
points(SOMoutput()$Coord$X[SLN[i]], SOMoutput()$Coord$Y[SLN[i]],pch=16,cex=1,col="black")
mtext(paste0("Qe=", round(SelDay()[,4][i],digit=3)),side=1,lin=0.5);
#--------qua parte destra con boxplot:
#-------------------senza whiskers!!!!--------------------------------
StepB<-1/nClus;
SeqB<-abs(seq(-1,0,StepB));
NULLO<-data.frame(matrix(-100000,nrow=nClus,ncol=ncol(SOMoutput()$codebook)))
NULLO[SelDay()[i,"Cluster"],]<-SelBOX()[i,-1]
for (w in c(1:nClus)) {
############################### CORRETTO DA QUI: 20210211 ###############################################
opar <- par(fig=c(0.5, 1, SeqB[w+1], SeqB[w]),mar=c(1,1,1,0.5),oma=c(0.5,0.5,0,0.5),xpd=FALSE,pty="m",family="serif");
############################### A QUI. ######################################################################
opar <- par(new=TRUE);
boxplot(SOMoutput()$codebook[which(clusNum4()==w),1],range=0,xaxt="n",yaxt="n",xlim=Xlim,ylim=Ylim,boxwex=0.9,at=1,whisklty = 0, staplelty = 0);
abline(h=pitch,col="gray");
for (f in c(1:ncol(SOMoutput()$codebook)))
{boxplot(SOMoutput()$codebook[which(clusNum4()==w),f],range=0,xaxt="n",yaxt="n",boxwex=1,at=f,add=TRUE,whisklty = 0, staplelty = 0)};
axis(1,at=seq(1,ncol(SOMoutput()$codebook),1),labels=NA,tcl=-0.3,cex.axis=Xdim);
############################### CORRETTO DA QUI: 20210211 ###############################################
axis(1,at=seq(1,ncol(SOMoutput()$codebook),1),labels=Names,lwd=0,line=-0.5,cex.axis=Xdim);
############################### A QUI. ######################################################################
axis(2,at=pitch,labels=NA,tcl=-0.3,cex.axis=0.6);
############################### CORRETTO DA QUI: 20210211 ###############################################
axis(2,at=pitch,labels=pitch,lwd=0,line=-0.7,cex.axis=0.6,las=2);
points(seq(1,ncol(SOMoutput()$codebook),1),NULLO[w,], pch=16,col="red",cex=0.7)
############################### A QUI. ######################################################################
on.exit(par(opar))
}
on.exit(par(opar))
} # END SingleSampPlot
output$plot62<- renderPlot({SingleSampPlot()})
output$down65 <- downloadHandler(filename =function() {paste0("Single sample",".pdf")},
content = function(file) {
pdf(file,height=input$h61,width=input$w61)
SingleSampPlot()
dev.off()}
)
#------------------------------------------------------------------------------------------------
session$onSessionEnded(function() {
stopApp()
})
} # server END
#---------------------------SHINY APP BUILT UP -------------------------------------
shinyApp(ui, server)
Try the SOMEnv package in your browser
Any scripts or data that you put into this service are public.
R Package Documentation
Browse R Packages
We want your feedback!
Note that we can't provide technical support on individual packages. You should contact the package authors for that.
Embedding an R snippet on your website
Add the following code to your website.
For more information on customizing the embed code, read Embedding Snippets.