R/markui.r

In berdaniera/markr: markr / Human-assisted anomaly detection

require(shiny)
markui = fluidPage(title="markr/",
  wellPanel(HTML(paste0(
    "<center><font style='font-family:sans-serif; font-size:45px'>markr</font> <font style='font-size:30px'>",icon("magic"),"</font></center>"
  ))),
  sidebarLayout(
  sidebarPanel(
    uiOutput("fileup"),
    strong("2. Training data:"),
    tabsetPanel(
      tabPanel("The uploaded data", value="new",
        # choose training data
        br(),
        # select columns to include in model
        selectizeInput("vari", "Choose variables for training model:", choices=NULL, multiple=TRUE),
        # enter filters for training data
        textInput("filt", "Enter data filter to apply for training data (optional):",placeholder="Temp<20&DateTime>'2016-01-01'"),
        HTML(paste0("<i>This is a conditional statement.</i> For help, ",actionLink("conditionals","click here."))),
        conditionalPanel("input.conditionals%2 == 1",
          "Need to add some description of conditional statements here."
        ),
        checkboxInput("delt", "Include the change in the variables as a feature in the anomaly detection?", value=TRUE),
        br()
      ),
      tabPanel("Previously-trained data", value="old",
        br(),
        selectizeInput("predat", "Choose existing flagged dataset:", choices="No data found...", multiple=FALSE)
      ),
    id="traindat",type="tabs"),
    # choose nu value for SVM model
    sliderInput("nu", "3. Select an error threshold for training data:", 0, 0.2, 0.01),
    p("Tip: As the training data are filtered to only include 'good' data, the error threshold can be lower (since we expect fewer errors in the training data)."),
    htmlOutput("fiterr"), # check if there is an error in model
    uiOutput("fitbutton"), # load model fit button if data is uploaded
    HTML("<br><center><font color='#666666'><i>Note: The model can take some time to fit... please be patient.</i></font></center>"),
    br(),
    actionLink("showeg","How does this work?"),
    conditionalPanel("input.showeg%2 == 1",
      HTML("The model is a One-class <a href='https://en.wikipedia.org/wiki/Support_vector_machine'>Support Vector Machine</a> with a radial kernel. It fits a decision boundary based on the multivariate relationship between the training data. In this example, the training data are shown as open circles. Randomly generated test data are overlaid in colors, based on whether the trained model classified them as 'good' or as 'anomalous'.<br><br>"),
      actionButton("egdata","Generate example training and test data"),
      checkboxInput("egshow","Show test data on graph"),
      plotOutput("egplot")
    ),
    br(),br(),
    HTML("<center><font color='#aaa'><i>Built by <a href='http://github.com/berdaniera/'>Aaron</a>. Powered by <a href='http://shiny.rstudio.com/'>Shiny.</a></i></font></center>")
  ),
  mainPanel(
    fluidRow(
      wellPanel(
        fluidRow(column(width=12,h4(actionLink("taginstructions","Click here for instructions")))),
        fluidRow(column(width=12,
          conditionalPanel("input.taginstructions%2 == 1",
            HTML("<ul>
              <li>Highlight data by dragging and selecting on the graph.</li>
              <li><b>Mark</b> data points by highlighting and clicking the buttons below.</li>
              <li><b>Store</b> flags by:<ol>
                <li><i>highlighting</i> the desired marked data,</li>
                <li><i>naming</i> the flag (and adding optional comments),</li>
                <li><i>clicking</i> 'Store selected flags' button.</li></ol>
              <li><b>Zoom</b> in on the graph by highlighting a time range and double clicking on the graph.</li>
              <li><b>When done:</b> Save and/or download the flagged data.</li>
            </ul>")
          )
        )),
        fluidRow(column(width=12,
          actionButton("flag_new", "Mark selected", icon=icon("flag"),
            style="color:#fff; background-color: #E69F00; border-color: #fff"),
          actionButton("flag_erase", "Un-mark selected", icon=icon("flag-o"),
            style="color:#fff; background-color: #CC79A7; border-color: #fff"),
          actionButton("flag_clear", "Clear unstored flags", icon=icon("eraser"),
            style="color: #fff; background-color: #009E73; border-color: #fff"),#d9534f
          actionButton("na_rm", "Remove NA values", icon=icon("times-circle"),
            style="color: #fff; background-color: #666666; border-color: #fff"),
          actionButton("flag_reset", "Reset zoom", icon=icon("repeat"),
            style="color: #fff; background-color: #ff7f50; border-color: #fff"),
          actionButton("flag_store", "Store selected flags", icon=icon("database"),
            style="color: #fff; background-color: #337ab7; border-color: #fff")
        )),br(),
        fluidRow(
          column(width=4,selectizeInput("flag_name", "Flag name/ID (either choose or type to add):", choices = NULL, options = list(create = TRUE))),
          column(width=8,textInput("flag_comments","Flag comment:",placeholder="Enter flag comments"))
        ),
        fluidRow(column(width=12,
          actionButton("flag_save", "Save flags and data for future fits", icon=icon("folder"),
            style="color: #fff; background-color: #337ab7; border-color: #fff"),
          actionButton("flag_download", "Download data and metadata as .csv", icon=icon("download")),
          actionButton("flag_return", "Return flags and model object in R", icon=icon("terminal"),
            style="color: #fff; background-color: #ff6a80; border-color: #fff")
        )),
        textOutput("loadtext")
      )
    ),
    fluidRow(column(width = 12,
        plotOutput("flagplot", dblclick = "plot_dblclick", brush = brushOpts(id = "plot_brush",direction="x"))
    ))
  )
  )#,
#  wellPanel(HTML("<center><font color='#aaa'><i>Built by <a href='http://github.com/berdaniera/'>Aaron</a>. Powered by <a href='http://shiny.rstudio.com/'>Shiny.</a></i></font></center>"))
)