inst/shiny/lawlarge.R

In gbonte/gbcode: Code from the handbook "Statistical foundations of machine learning"

library(shinydashboard)
library(mvtnorm)
library(scatterplot3d)
library(ellipse)
library(plot3D)
library(markdown)
BOUND1<-5
BOUND2<-2
ui <- dashboardPage(
  dashboardHeader(title="InfoF422: Frequency"),
  dashboardSidebar(
    sidebarMenu(
      sliderInput("N",
                  "Number of samples:",
                  min = 2,
                  max = 50000,
                  value = 50,step=2),
      
      sliderInput("P",
                  "Probability of success:",
                  min = 0,
                  max = 1,
                  value = 0.5,step=0.05),
      actionButton("sample", label = "Sample"),
      checkboxInput("Normalized", label = "Normalized", value = FALSE),
      sliderInput("Yrange",
                  "Range Y:",
                  min = 0,
                  max = 1,
                  value = c(0,1),step=0.01),
      sliderInput("Xrange",
                  "Range X:",
                  min = 0,
                  max = 100,
                  value = c(0,100),step=1),
      
      menuItem("Large numbers", tabName = "Pvalue", icon = icon("th")),
      menuItem("About", tabName = "about", icon = icon("question"))
    ),
    sidebarMenuOutput("menu")
  ),
  dashboardBody(
    tabItems(
      #
      tabItem(tabName = "Pvalue",
              
              
              fluidRow(   
                box(width=8,title = "Frequency of success",plotOutput("frequency"))),
              fluidRow( box(width=8,title = "(number successes) - (number failures)",plotOutput("headstails"))
              #  valueBoxOutput("Statistic")
             
              )
      ),
      tabItem(tabName = "about",
              fluidPage(
                includeHTML("about/about.lawlarge.html")
              )
      )
    )
  )
) # ui

d<-NULL
ns<-NULL
BOUND<-150
sX<-NULL
require(statmod)
server<-function(input, output,session) {
  
  
  
  observeEvent(input$N,  {
    updateSliderInput(session = session, inputId = "Xrange", max = input$N, value = c(input$Xrange[1],input$N))
  })
  
  
  output$frequency = renderPlot({
    
    N<-input$N
    input$sample
    X<-rbinom(N,1,prob=input$P) ##sample(c(0,1),N,prob=c(1-input$P,input$P),rep=TRUE)
    sX<<-cumsum(X)
    
    
    f<-NULL
    d<<-NULL
    ns<<-NULL
    for (i in 1:length(sX)){
      f<-c(f,sX[i]/i)
      d<<-c(d,2*sX[i]-i) ## # successes= sum(X[1:i]), # fails= i- sum(X[1:i])
      ns<<-c(ns,(2*sX[i]-i)/i)
    }
    phat=f[length(f)]
    alpha=0.01
    eps=qnorm(1-alpha/2)*sqrt(input$N*phat*(1-phat))/input$N
      
    plot(1:(length(f)),f, type="l",ylim=input$Yrange,xlab="Number of trials",ylab="Frequency",xlim=input$Xrange)
    lines(1:(length(f)),numeric(length(f))+input$P,col="red")
    lines(1:(length(f)),numeric(length(f))+phat+eps,col="green")
    lines(1:(length(f)),numeric(length(f))+phat-eps,col="green")
    
  })
  
  
  output$headstails = renderPlot({
    input$sample
   input$N
    input$P
    if (!input$Normalized) {
      plot(d,ylim=c(min(d),max(d)),type="l",xlim=input$Xrange)
    }else {
      plot(1:input$N,ns,ylim=c(-1,1),type="l",xlim=input$Xrange)
      lines(1:input$N,numeric(input$N)+2*input$P-1,col="red") 
    }
  })
 
  
}



shinyApp(ui, server)