K-means clustering interactive tutorial

server.R

library(ggplot2)
library(shiny)

shinyServer(function(input, output) {
    output$main_plot <- renderPlot({
        set.seed(205)
        means = t(replicate(input$clusters, c(runif(1), runif(1))))
        means = cbind(1:input$clusters, means)
        means = apply(means, 2, function(col) rep(col, input$points))
        means = data.frame(means)
        colnames(means) = c("Cluster", "cX", "cY")
        means$Cluster = factor(means$Cluster)
        means$X = rnorm(NROW(means), means$cX, input$sd)
        means$Y = rnorm(NROW(means), means$cY, input$sd)

        if (input$iterations > 0) {
            k = kmeans(as.matrix(means[c("X", "Y")]), input$kclusters,
                            iter.max=input$iterations)
            means$Cluster = factor(k$cluster)
        }
    
        g = ggplot(means, aes(X, Y)) + geom_point(aes(col=Cluster))
        if (input$iterations > 0) {
            g = g + geom_point(aes(X, Y), data=as.data.frame(k$centers),
                                pch="X", size=10)
        }
        print(g)
    })
})

ui.R

library(shiny)

shinyUI(bootstrapPage(
  sliderInput(inputId = "clusters",
      label = "Number of true clusters in data",
      min=1, max=6, value=3),

  sliderInput(inputId = "sd",
      label = "Standard deviation of clusters",
      min=.001, max=.2, value=.02),

  sliderInput(inputId = "kclusters",
      label = "Number of clusters used in the k-means algorithm",
      min=1, max=6, value=3),

  sliderInput(inputId = "points",
      label = "Number of points per cluster",
      min=3, max=100, value=25),  

  sliderInput(inputId = "iterations",
      label = "Number of iterations of k-means algorithm",
      min=0, max=6, value=3),

  plotOutput(outputId = "main_plot", height = "300px")
  #plotOutput(outputId = "wss_plot", height = "300px")
))