inst/app/app.R

In AngelosPsy/condir: Computation of P Values and Bayes Factors for Conditioning Data

# Shiny Application for condir
library(shiny)
library(shinythemes)
library(BayesFactor)
library(condir)
library(foreign)
library(tools)

# Define UI for application that draws a histogram
ui <- shiny::shinyUI(
      shiny::fluidPage(theme = shinytheme("flatly"),
      shiny::titlePanel("condir"),
      shiny::sidebarLayout(
      shiny::sidebarPanel(
      shiny::fileInput("file", label = h4("Choose data file to analyse")),
      tags$hr(),
      shiny::checkboxInput('header', 'Header', TRUE),
      shiny::radioButtons('sep', 'Separator',
                   c(Tab='\t', Comma=',', Semicolon=';'),
                   selected = '\t'),
      shiny::radioButtons('quote', 'Quote',
                   c(None='', 'Double Quote'='"',
                     'Single Quote'="'"), selected = '"'),
      tags$hr(),
      p("Accepted file extensions: .txt, .csv, .sav.
         Uploading any other type of file will result in an error."
      ), width = 3
    ),
    # Main panel
    shiny::mainPanel(
      shiny::tabsetPanel(
        shiny::tabPanel("Data", shiny::dataTableOutput("contents")),
        shiny::tabPanel("Results",
                        h4("Descriptive results"),
                        shiny::tableOutput("desc"),
                        h4("Frequentist Results"),
                        shiny::tableOutput("freq"),
                        h4("Bayesian Results"),
                        shiny::tableOutput("bayes"),
                        shiny::uiOutput("cs1"), shiny::uiOutput("cs2"),
                        shiny::uiOutput("group"),
                        shiny::uiOutput("alternative"),
                        shiny::uiOutput("Confidence Level"),
                        shiny::uiOutput("mu"),
                        shiny::textInput(inputId = "rscale",
                               label = "What is the width of the Cauchy prior?",
                                         value = ".707"),
                        shiny::textInput(inputId = "sigLevel",
                                         label = "What is the alpha level?",
                                         value = "0.05")),
        shiny::tabPanel("Plots", shiny::plotOutput("plot", width = "100%"),
                        shiny::uiOutput("bfChoice"),
                        shiny::plotOutput("robplot")),
        shiny::tabPanel("Summary",
                        h4("Summary of main results"),
                        shiny::verbatimTextOutput("summaryMain"),
                        h4("Summary of robustness test"),
                        shiny::verbatimTextOutput("summaryRob"),
                        shiny::radioButtons(inputId = "interpretation",
                                  label = "Should the result be interpreted?",
                                  choices = c("TRUE", "FALSE"),
                                  selected = "FALSE", inline = TRUE)),
        shiny::tabPanel("Help",
                        h3("General Description"),
                        p("This Shiny application was designed so as to enable the
                          easier use of the", code("condir"), "package. Step-to-step examples
                          on how to use. An example of how to use", code("condir"), "via the R console
                          are available here: ", a("https://github.com/AngelosPsy/condir"),
                          "An example on how to use ", code("condir"), "can be found in Krypotos,
                          Klugkist, and Engelhard (under revision).

                          For bug and feature requests, please use the relevant github page",
                          a("https://github.com/AngelosPsy/condir/issues"), "."),

                        h3("Explanation of each tab"),
                        p(code("condir"), "has 5 different tabs, each one with each function.
                          Here, we provide brief explanations of each tab and
                          we explain how data can be uploaded in ", code("condir"), "." ),
                        h4("Data input"),
                        p("By using the Browse button located on the left of the screen, the user
                          can search and upload the data to be analysed. The current version of ",
                          code("condir"), "accepts data files with the 'txt', 'csv', and 'sav' extensions. Note
                          that the used data input method is based on the example used in Rstudio and that it
                          can be find here: ", a("http://shiny.rstudio.com/gallery/upload-file.html")),
                        h4("Data Tab"),
                        p("After uploading the data, the Data tab will present the uploaded data. Please note
                         that if the data columns use another than the 'tab' separator, the data will not be presented
                         in separate column but into a single one. To change that, select the correct separator on the left
                         of the tab. A sceenshot of the data panel follows: "),
                        img(src='data.png', align = "center", width="600", height="400"),
                        h4("Results Tab"),
                        p("The results tab allows the user to select which column has the CS1, the CS2, and the group (optional)
                          data. After all these options have been made, the Frequentists and the Bayesian results are returned. On
                          the bottom of this page, the user can also control a few more analyses options (e.g., the scale factor
                          of the Cauchy distribution). The revelant screenshot follows: "),
                        img(src='results.png', align = "center", width="600", height="400"),
                        h4("Plots Tab"),
                        p("The plots tab depicts the mean responses for each stimulus -- and for each group if a between group
                          analyses was performed. On the bottom, the relevant robustness test is depicted. This plot is based on
                          a similar plot provided in ", a("https://jasp-stats.org/"), "For a screenshot of this tab, see below:"),
                        img(src='plots.png', align = "center", width="600", height="400"),
                        h4("Summary Tab"),
                        p("The summary of the results is provided in this tab. The user may choose to interpret the findings.
                          The interpretation is based on the numerical values of the results, and the user is free to accept or ignore
                          the provided interpretation. The relevant screenshot follows: "),
                        img(src='summary.png', align = "center", width="600", height="400"),
                        h3("Reference"),
                         ("Krypotos, A-M., Klugkist, I., & Engelhard, I. M. (under revision). Bayesian Hypothesis Testing for Human Threat Conditioning Research: An introduction and the
                         condir R package."))
        ) # tabsetPanel
   ) # mainPanel
  ) # sidebarLayout
) # fluidPage
) # shinyUI

server <- shiny::shinyServer(function(input, output, session) {
  session$onSessionEnded(stopApp)
  datz <- shiny::reactive({
          shiny::req(input$file)
          inFile <- input$file

          if (is.null(inFile)) return(NULL)
            ext <- tools::file_ext(inFile$name)
          if(ext == "txt"){
            datz <- utils::read.table(inFile$datapath, header = input$header,
                               sep = input$sep, quote = input$quote)
          } else if(ext == "csv"){
            datz <- utils::read.csv(inFile$datapath, header = input$header,
                               sep = input$sep, quote = input$quote)
          } else if(ext == "sav"){
            datz <- foreign::read.spss(file = inFile$datapath,
                                       to.data.frame = TRUE)
          }
         data.frame(datz)
  })

  # Data tab
  output$contents <- shiny::renderDataTable(expr = datz(),
                                     options = list(lengthMenu = c(5, 30, 50),
                                                    pageLength = 5))

  colNames <- shiny::reactive({colnames(datz())})

  # Results tab
  output$cs1 <-  shiny::renderUI({
    selectInput("cs1Button", label = "Which column has the CS1 data?",
              choices = c("", colNames()),
              selected = "")
  })

  cs1Ch <- shiny::reactive({
    shiny::req(input$cs1Button)
    datz()[, colnames(datz()) %in% input$cs1Button]
  })

  # CS2
  output$cs2 <-  shiny::renderUI({
    selectInput("cs2Button", label = "Which column has the CS2 data?",
                choices = c("", colNames()),
                selected = "")
  })

  cs2Ch <- shiny::reactive({
    shiny::req(input$cs2Button)
    datz()[, colnames(datz()) %in% input$cs2Button]})

  # Group
  output$group <-  shiny::renderUI({
    selectInput("groupButton", label = "Which column has the group data? (Optional)",
                choices = c("NULL", colNames()),
                selected = 1)
  })


  groupCh <- shiny::reactive({
    shiny::req(input$groupButton)
    datz()[, colnames(datz()) %in% input$groupButton]
  })

  groupPress <- shiny::reactive({
    shiny::req(input$groupButton)
    input$groupButton
  })

  # Alternative
  output$alternative <- shiny::renderUI({
                        shiny::radioButtons(inputId = "alternativeButton",
                        label = "What is the alternative hypothesis?",
                        choices = c("two.sided", "greater", "less"),
                        inline = TRUE)
  })

  alternativeCh <- shiny::reactive({input$alternativeButton})

  # rscale
  rscale <- shiny::reactive({input$rscale})

  csComp <- shiny::reactive({
                   condir::csCompare(cs1 = cs1Ch(), cs2 = cs2Ch(),
                                     group = groupCh(),
                                     alternative = alternativeCh(),
                                     rscale = rscale(),
                                     boxplot = FALSE)
  })

  csSens <- shiny::reactive({
                   condir::csSensitivity(cs1 = cs1Ch(), cs2 = cs2Ch(),
                           rscaleSens = c(.1, .3, .5, .7, 1, 1.5),
                           group = groupCh())
  })

  # Plot tab
  output$plot <- shiny::renderPlot({
                 condir::csPlot(cs1 = cs1Ch(), cs2 = cs2Ch(), group = groupCh())
  }, width = 600)

  # Robustness Plot tab
  output$bfChoice <- shiny::renderUI({
                            shiny::req(input$file)
                            shiny::radioButtons(inputId = "bfChoice",
                            label = "What should be plotted?",
                            choices = c("BF01", "BF10"), selected = "BF01",
                            inline = TRUE)
  })

  robChoice <- shiny::reactive({input$bfChoice})

  intChoice <- shiny::reactive({input$interpretation})

  selSigLevel <- reactive({input$sigLevel})

  output$robplot <- shiny::renderPlot({
                           BF01 <- ifelse (robChoice() == "BF01", TRUE, FALSE)
                           condir::csRobustnessPlot(cs1 = cs1Ch(),
                                   cs2 = cs2Ch(), group = groupCh(),
                                   alternative = alternativeCh(),
                                   data = NULL,
                                   rscaleSens = c(.1, .3, .5, .7, 1, 1.5),
                                   BF01 = BF01)}, width = 600)

  # Results tab
  output$desc <- shiny::renderTable({
      if (groupPress() != "NULL"){
        tmp <- do.call("rbind", csComp()$descriptives)
      } else {
        tmp <- csComp()$descriptives
      }
     tmp$vars <- row.names(tmp)
     data.frame(tmp)
     }, caption.placement = getOption("xtable.caption.placement", "top")
    )

  output$freq <- shiny::renderTable({
    csComp()$freq.results

    }, caption.placement = getOption("xtable.caption.placement", "top")
  )

  output$bayes <- shiny::renderTable({
    csComp()$bayes.results},
    caption.placement = getOption("xtable.caption.placement", "top")
    )

  # Summary tab
  output$summaryMain <- shiny::renderPrint({
                               shiny::req(input$file)
                               condir::csReport(csCompareObj = csComp(),
                                                alphaLevel = selSigLevel(),
                                                interpretation = intChoice())
  })

  output$summaryRob <- shiny::renderPrint({
                            shiny::req(input$file)
                            condir::csReport(csSensitivityObj = csSens(),
                                             alphaLevel = selSigLevel(),
                                             interpretation = intChoice())
   })
}) # shinyServer

# Run the application
shinyApp(ui = ui, server = server)