Problem passing variable names via selectInput() in R/Shiny

Question

I'm building a shinyapp where I am passing variable names as arguments to the server with the selectInput() widget. The following static example illustrates the plot that I want displayed:

  g <- ggplot(d, aes(y, fill = var1, colour = var1)) + 
    geom_density(alpha=.2) 

Just to be clear, here is an image of the above plot

What I want to do in my shinyapp is to separate the plotted distributions by different variables that the user can choose via selectInput(). But when I substitute var1 in the above with a generic argument, this is not what I get. Please see the example:

library(ggplot2)
library(shiny)

d <- data.frame(var1=as.factor(sample(x = 1:4, size = 250, replace = TRUE)),
                var2=as.factor(sample(x = 1:4, size = 250, replace = TRUE)),
                y=rnorm(250, mean = 0, sd = 1))

nms <- names(d)

ui <- fluidPage(selectInput(inputId ="var", "Variable:",
                            choices = nms, selected = "var1"),
                plotOutput(outputId = "plot")
)

server <- function(input, output) {

  g <- ggplot(d, aes(y, fill = input$var, colour = input$var)) + 
    geom_density(alpha=.2) 


  output$plot <- renderPlot(g)

}

shinyApp(ui,server)

The actual output I get is different from the static example I started with (click through for image).

Can someone please point out my error? Thankful for help.

Answer 1

input$var is a string. Therefore, do

output$plot <- renderPlot({
  g <- ggplot(d, aes_string("y", fill = input$var, colour = input$var)) + 
         geom_density(alpha=.2)
  g
})