R/Shiny: Change plot ONLY after action button has been clicked - r

Question

I am setting up a small shiny app where I do not want the plot to change unless the action button is clicked. In the example below, when I first run the app, there is no plot until I click the action button. However, if I then change my menu option in the drop-down from Histogram to Scatter, the scatter plot is automatically displayed even though the value for input$show_plot has not changed because the action button has not been clicked.
Is there a way that I can change my menu selection from Histogram to Scatter, but NOT have the plot change until I click the action button? I've read through several different posts and articles and can't seem to get this worked out.
Thanks for any input!
ui.R
library(shiny)
fluidPage(
tabsetPanel(
tabPanel("Main",
headerPanel(""),
sidebarPanel(
selectInput('plot_type', 'Select plot type', c('Histogram','Scatter'), width = "250px"),
actionButton('show_plot',"Plot", width = "125px"),
width = 2
),
mainPanel(
conditionalPanel(
"input.plot_type == 'Histogram'",
plotOutput('plot_histogram')
),
conditionalPanel(
"input.plot_type == 'Scatter'",
plotOutput('plot_scatter')
)
))
)
)
server.R
library(shiny)
library(ggplot2)
set.seed(10)
function(input, output, session) {
### GENERATE SOME DATA ###
source_data <- reactive({
mydata1 = as.data.frame(rnorm(n = 100))
mydata2 = as.data.frame(rnorm(n = 100))
mydata = cbind(mydata1, mydata2)
colnames(mydata) <- c("value1","value2")
return(mydata)
})
# get a subset of the data for the histogram
hist_data <- reactive({
data_sub = as.data.frame(source_data()[sample(1:nrow(source_data()), 75), "value1"])
colnames(data_sub) <- "value1"
return(data_sub)
})
# get a subset of the data for the scatter plot
scatter_data <- reactive({
data_sub = as.data.frame(source_data()[sample(1:nrow(source_data()), 75),])
return(data_sub)
})
### MAKE SOME PLOTS ###
observeEvent(input$show_plot,{
output$plot_histogram <- renderPlot({
isolate({
plot_data = hist_data()
print(head(plot_data))
p = ggplot(plot_data, aes(x = value1, y = ..count..)) + geom_histogram()
return(p)
})
})
})
observeEvent(input$show_plot,{
output$plot_scatter <- renderPlot({
isolate({
plot_data = scatter_data()
print(head(plot_data))
p = ggplot(plot_data, aes(x = value1, y = value2)) + geom_point()
return(p)
})
})
})
}

Answer 1

Based on your desired behavior I don't see a need for actionButton() at all. If you want to change plots based on user input then the combo of selectinput() and conditionPanel() already does that for you.
On another note, it is not good practice to have output bindings inside any reactives. Here's an improved version of your server code. I think you are good enough to see notice the changes but comment if you have any questions. -
function(input, output, session) {
### GENERATE SOME DATA ###
source_data <- data.frame(value1 = rnorm(n = 100), value2 = rnorm(n = 100))
# get a subset of the data for the histogram
hist_data <- reactive({
# reactive is not needed if no user input is used for creating this data
source_data[sample(1:nrow(source_data), 75), "value1", drop = F]
})
# get a subset of the data for the histogram
scatter_data <- reactive({
# reactive is not needed if no user input is used for creating this data
source_data[sample(1:nrow(source_data), 75), , drop = F]
})
### MAKE SOME PLOTS ###
output$plot_histogram <- renderPlot({
req(hist_data())
print(head(hist_data()))
p = ggplot(hist_data(), aes(x = value1, y = ..count..)) + geom_histogram()
return(p)
})
output$plot_scatter <- renderPlot({
req(scatter_data())
print(head(scatter_data()))
p = ggplot(scatter_data(), aes(x = value1, y = value2)) + geom_point()
return(p)
})
}