Data Quality Explorer Prototype

app.R

library(shiny)
library(plotly)
library(dplyr)

data(diamonds, package = "ggplot2")
df <- as_tibble(diamonds)
vars <- names(df)
numeric_vars <- df %>% select_if(is.numeric) %>% names()
categorical_vars <- df %>% select_if(~ !is.numeric(.x)) %>% names()

ui <- fluidPage(
  headerPanel("Data Quality Explorer"),
  
  # Inputs 
  sidebarPanel(
    sliderInput('sample_size', 'Sample Size', min = 1, max = nrow(df), value = 30000, step = 500, round = 0),
    selectInput('x', 'Variable', choices = numeric_vars, selected = "price"),
    sliderInput('bins', 'Histogram Bins', min = 10, max = 200, value = 50),
    sliderInput('plot_height', 'Height of plot (in pixels)', min = 100, max = 1500, value = 750),
    selectInput('facet', 'Facet', categorical_vars, selected = "time_sk")
  ),
  # UI Outputs
  mainPanel(
    tabsetPanel(
      type = "tabs",
      tabPanel("Variable Summary", uiOutput("dt")),
      tabPanel("Histogram",        plotlyOutput('histogram')),
      tabPanel("Density Plot",     plotlyOutput('density_plot'))
    )
  )
)

server <- function(input, output) {
  
  dataset <- reactive({
    df[sample(nrow(diamonds), input$sample_size),]
  })
  
  summaries <- df %>% mutate_if(is.integer, as.numeric) %>% skimr::skim_to_list()
  summary_nms <- names(summaries)
  
  observe({
    lapply(summary_nms, function(summary) {
      output[[paste0('summary_', summary)]] <- DT::renderDT(
        summaries[[summary]],
        style = "bootstrap",
        options = list(pageLength = 20, lengthChange = FALSE))
    })
  })

  output$dt <- renderUI({
    tagList(lapply(summary_nms, function(summary) {
      DT::dataTableOutput(paste0('summary_', summary))
    }))
  })
  
  output$histogram <- renderPlotly({
    
    p <- ggplot(dataset(), aes_string(input$x, fill = input$facet)) + 
      geom_histogram(bins = input$bins) + 
      theme_minimal() + 
      facet_wrap(input$facet, ncol = 1, scales = "free")
    
    ggplotly(p, height = input$plot_height) %>% 
      layout(autosize=TRUE)
  })
  
  output$density_plot <- renderPlotly({
    
    # build graph with ggplot syntax
    p <- ggplot(dataset(), aes_string(input$x, fill = input$facet)) + 
      geom_density() +
      theme_minimal() + 
      theme(axis.text.y = element_blank()) +
      facet_wrap(input$facet, ncol = 1, scales = "free")
    
    # if (input$facet_row) p <- p + facet_grid(input$facet_row)
    
    ggplotly(p, height = input$plot_height) %>% 
      layout(autosize=TRUE)
  })
  
}

shinyApp(ui, server)