Using lead/lag in a Shiny App

Question

I feel like this question applies to a general query about how to manipulate data within a Shiny context but my specific question relates to the usage of the lead/lag functions in dplyr. Right now I am faced with two problems.

The first problem is defining the variables in the dataset that have been lead/lag. I am using a dplyr approach where I name the lagged variable yvar. CLearly this isn't correct as R isn't able to find yvar. So can anyone recommend a way to change/create a variable so that the result is transferred down through the shiny app and available to be called?

The second problem is specifying the number of rows to lead/lag the Y variable. It is easy enough in the ui.R file to specify an input box with lead/lag positions. But then how do I place the integer from that input box as an argument in the lead function?

Here is the ui.R:

library(shiny)
library(ggplot2)
library(dplyr)


dataset <- iris

shinyUI(pageWithSidebar(

  headerPanel("Iris Data Explorer"),

  sidebarPanel(


    selectInput('x', 'X', names(dataset), names(dataset)[[2]]),
    selectInput('y', 'Y', names(dataset), names(dataset)[[4]]),
    selectInput('offset', 'Offset Level of Y', 0:5, 0),
    selectInput('species', 'Species', levels(dataset$Species), "virginica")
  ),

  mainPanel(
    plotOutput('plot')
  )
))

And here is the server.R:

library(shiny)
library(ggplot2)
library(dplyr)


shinyServer(function(input, output) {

  dataset <- reactive({
    iris %>%
      filter(Species==input$species) 
    #%>% mutate(yvar=lead(input$y, paste0(input$offset)))
  })



  output$plot <- renderPlot({


    dataset<-dataset()
    p <- ggplot(dataset, aes_string(x=input$x,
                                    #y=yvar)
                                    y=input$y
    )) +
      geom_point() 

    print(p)

  }, height=500)

})

score 3 · Accepted Answer · edited May 23 '17 at 11:50

3

I think this is what you want. To get it to work I did the following:

abandoned mutate because it apparently is quite difficult to use dynamic names with it (see:R - dplyr - mutate - use dynamic variable names) and just did it with data frame indexing as suggested there.
combined it into one file so I can see everything at once (good for small stuff)
added a print statement to dump the dataset on every change so you can see what is happening.
added quotes to the yvar variable so that it works with aesstring
added an as.numeric to the offset variable as an input to lead

This is the resulting code:

library(shiny)
library(ggplot2)
library(dplyr)

dataset <- iris

u <- shinyUI(pageWithSidebar(
  headerPanel("Iris Data Explorer"),
  sidebarPanel(
    selectInput('x', 'X', names(dataset), names(dataset)[[2]]),
    selectInput('y', 'Y', names(dataset), names(dataset)[[4]]),
    selectInput('offset', 'Offset Level of Y', 0:5, 0),
    selectInput('species', 'Species', levels(dataset$Species), "virginica")
  ),
  mainPanel(
    plotOutput('plot')
  )
))
s <- shinyServer(function(input, output) {
  dataset <- reactive({
    df <- iris %>% filter(Species==input$species) 
    df[["yvar"]] <- lead(df[[input$y]],as.numeric(input$offset))
    return(df)
  })
  output$plot <- renderPlot({
    dataset<-dataset()
    print(dataset)
    ggplot(dataset, aes_string(x=input$x,y="yvar")) + geom_point(na.rm=T) 
  }, height=500)
})
shinyApp(ui=u,server=s)

And this is what it looks like:

edited May 23 '17 at 11:50

Community

1
1

answered Mar 16 '16 at 23:02

Mike Wise

22,131
8
81
104

Actually using `mutate` with dynamic names is trivial — just switch to `mutate_` and provide a named vector with the transformations. – Konrad Rudolph Mar 16 '16 at 23:08
Then that link (that Hadley was on) is obsolete? So how should it look? Your description is too brief for me. – Mike Wise Mar 16 '16 at 23:11
Not obsolete, just incomplete. OP’ code, using `mutate_`, would look as follows: `mutate_(yvar = bquote(lead(.(as.name(y)), .(as.numeric(offset)))))` — in general, just pass an expression or formula, with the desired names injected. I myself am partial to `bquote` but dplyr also comes with `lazyeval` which makes constructing these expressions handy (shown in the linked answer). – Konrad Rudolph Mar 16 '16 at 23:29
Thanks. The OP can decide to use that if they want. Someone (you perhaps?) maybe needs to add a new answer to that other post as well. – Mike Wise Mar 16 '16 at 23:32
I think it is probably best to go with the answer provided by here. `lazyeval` can be a bit confusing. Just so I understand thought `input$x` doesn't need quotes in `aes_string` because it is already a string? That is the format here? – boshek Mar 17 '16 at 00:14

Using lead/lag in a Shiny App

1 Answers1