R-shiny: Discrete value supplied to continuous scale - r

Question

So I am using mpg dataset to practice my R-shiny skills, but I encountered a problem.
I want to write a app which I could choose different variables to make graph, if it involves at least one discrete variable, then I draw a geom_boxplot, else, I will just draw a geom_point.
My ui.R looks like this:
library(shiny)
shinyUI(fluidPage(
# Application title
titlePanel("Old Faithful Geyser Data"),
# Sidebar with a slider input for number of bins
sidebarLayout(
sidebarPanel(
selectInput(inputId = "var1",
label = "Choose x variable",
choices =
names(mpg)
),
selectInput(inputId = "var2",
label = "Choose y variable",
choices =
names(mpg))
),
# Show a plot of the generated distribution
mainPanel(
plotOutput("distPlot")
)
)
))
And
My server.R looks like this:
library(shiny)
library(tidyverse)
library(dplyr)
shinyServer(function(input, output) {
output$distPlot <- renderPlot({
if(typeof(mpg[, input$var1]) == "chr" ||
typeof(mpg[, input$var2]) == "chr")
{
ggplot(data = mpg) +
geom_boxplot(mapping =
aes(x = mpg[, input$var1],
y = mpg[, input$var2])) +
xlab(input$var1) +
ylab(input$var2) +
ggtitle(paste("Plot", input$var1, "vs", input$var2))
}
else
{
ggplot(data = mpg) +
geom_point(mapping =
aes(x = mpg[, input$var1],
y = mpg[, input$var2])) +
xlab(input$var1) +
ylab(input$var2) +
ggtitle(paste("Plot", input$var1, "vs", input$var2))
}
})
})
But it simply does not work! When I choose discrete variable, it will return me an error message says "Discrete value applied to continuous scale". However, if I both choose continuous value, it seems to be work fine.
Why it shows such an error message like that?
Please help me!
Thank you very much!

Answer 1

You have a couple of issues here all on the server side:
One: If you check an example of your typeof you'll see that you always get a list:
typeof(mpg[,"model"])
#[1] "list"
typeof(mpg[,"displ"])
#[1] "list"
This is because [ doesn't extract the actual element, but the list containing that element. From ?"[":
Indexing by [ is similar to atomic vectors and selects a list of the specified element(s).
Rather you should use typeof(mpg[[input$var1]]), etc because you want to extract the element of the list (not the list that contains the element).
Two:
There is actually a specific function in ggplot, aes_string which selects the columns to plot based on a string.
Incorporating these two changes should make your shiny app work. I also simplified you server a bit to get rid of the common ggplot code.
server <- function(input,output){
output$distPlot <- renderPlot({
p <- ggplot(mpg) + xlab(input$var1) +
ylab(input$var2) +
ggtitle(paste("Plot", input$var1, "vs", input$var2))
if(typeof(mpg[[input$var1]]) == "character" |
typeof(mpg[[input$var2]]) == "character")
{
p <- p + geom_boxplot(mapping =
aes_string(x = input$var1,
y = input$var2))
}
else
{
p <- p + geom_point(mapping =
aes_string(x = input$var1,
y = input$var2))
}
return(p)
})
}