How to make mutate work with reactive on numerical column names? - r

Question

I don't really know to hook up mutate with reactive when dealing with names of columns that are numerical.
I've got data that looks something like this:
df <- tibble(a=c("a", "b", "c"), `1990`=c(1,2,3), `2010`=c(3,2,1))
Everything looks fine when I do:
p <- df %>%
mutate(newvar = `1990`)
But I want to use it in my Shiny App such that newvar is assigned to the input from select list.
I created this reactive for that;
selectedyear <- reactive({
input$select
})
But now it doesn't seem to work:
p <- df %>%
mutate(newvar = selectedyear())
I tried different modifications, like:
p <- df %>%
mutate(newvar = `selectedyear()`)
but nothing seems to work for me.
The full code of the app:
library(shiny)
library(tibble)
library(dplyr)
df <- tibble(a=c("a", "b", "c"), `1990`=c(1,2,3), `2010`=c(3,2,1))
ui <- fluidPage(
selectInput("select", "Select:", c(1990, 2010)),
tableOutput("val")
)
server <- function(input, output) {
selectedyear <- reactive({
input$select
})
output$val <- renderTable({
p <- df %>%
mutate(temperature = selectedyear())
p
})
}
shinyApp(ui, server)
The same thing, but with characters as input is easy. Do you know some hack around this?

Answer 1

You can do
output$val <- renderTable({
p <- df %>%
mutate(temperature = !!selectedyear())
p
})

Answer 2

The problem seems to be the non-syntactic name of the columns 1990, 2010. The approach below should work. It uses eval/parse inside the right hand side of the mutate call. Note that you usually do not need to create reactives of (and with only) inputs, because each input is already reactive.
library(shiny)
library(tibble)
library(dplyr)
df <- tibble(a=c("a", "b", "c"), `1990`=c(1,2,3), `2010`=c(3,2,1))
ui <- fluidPage(
selectInput("select", "Select:", c(1990, 2010)),
tableOutput("val")
)
server <- function(input, output) {
output$val <- renderTable({
df %>%
mutate(temperature = eval(parse(text = paste0("`", input$select,"`"))))
})
}
shinyApp(ui, server)
If you are only interested in the temperature column and you do not need to show the rest of the data, then dplyr::select lets you access the input$select variable in a much more straightforward fashion:
output$val <- renderTable({
df %>%
select(temperature = input$select)
})