Shiny: passing input$var to aes() in ggplot2
I am trying to write a Shiny app graphing the density of a variable VAL, by categories of age (AGE) or sex (SEX). The user selects "SEX" or "AGE" in the dropdown menu, and I have been trying to use fill = input$Group_select
in ggplot and ggvis.
In ggplot:
output$plot <- renderPlot(ggplot(gdat[gdat$GFR==input$GFR_select,]) +
aes(x= VAL, fill=input$Group_select) +
geom_density(adjust=input$slider1))
EDIT: as docendo pointed out, this can be fixed for ggplot using aes_string:
output$plot <- renderPlot(ggplot(gdat[gdat$GFR==input$GFR_select,]) +
aes(x= VAL) +
geom_density(adjust=input$slider1, aes_string(fill=input$Group_select))):
In ggvis:
gvis <- reactive(subset(gdat, GFR==input$GFR_select) %>%
ggvis(x = ~VAL) %>% group_by_(input$Group_select) %>%
layer_densities(adjust = input$slider1) %>%
add_axis("y", title = "Density", ticks="none") %>%
scale_numeric("x", domain = c(0, 230)) )
gvis %>% bind_shiny("ggvis", "ggvis_ui")
SOLUTION: using as.name(input$Group_select) will render the graph properly!
This is (was) what is rendered: Imgur link to shiny output. Interestingly, it seems that group_by_ correctly interprets input$Group_select, whereas input$Group_select is treated as a constant in fill=input$Group_select
Any ideas on how I could get the plots to render correctly?
Here is the full Shiny code:
ui.R
library(shiny)
library(ggplot2)
library(dplyr)
library(ggvis)
shinyUI(fluidPage(
titlePanel("eGFR in the ARIC study"),
sidebarPanel(
selectInput("Group_select",
label=h3("By-Variable"),
choices=c("AGE","SEX","ALL"),
selected="SEX"),
selectInput("GFR_select",
label=h3("Creatinine Measure"),
choices=c("BOTH", "CREATININE", "CYSTATIN", "MDRD"),
selected="MDRD"),
sliderInput("slider1", label = h3("Bandwith Adjustment"),
min = 0.5, max = 4, value = 1)
),
mainPanel(
uiOutput("ggvis_ui"),
ggvisOutput("ggvis"),
plotOutput("plot")
)
))
server.R
library(shiny)
source("helpers.R")
shinyServer(function(input, output) {
gvis <- reactive(subset(gdat, GFR==input$GFR_select) %>%
ggvis(x = ~VAL, fill = ~input$Group_select) %>% group_by_(input$Group_select) %>%
layer_densities(adjust = input$slider1) %>%
add_axis("y", title = "Density", ticks="none") %>%
scale_numeric("x", domain = c(0, 230)) )
gvis %>% bind_shiny("ggvis", "ggvis_ui")
output$plot <- renderPlot(ggplot(gdat[gdat$GFR==input$GFR_select,]) +
aes(x= VAL, fill=input$Group_select) +
geom_density(adjust=input$slider1))
})
Solution 1:
I'm not sure this is still an open question, but an alternative to the aes_string
would be to convert the string to a symbol:
library(ggplot2)
# usually you can put all the shared aesthetics in the first line
ggplot(mtcars, aes(mpg, hp, colour = cyl)) +
geom_point()
# when the input is a string, you can use aes_string(), but you'd
# have to enter the string var separately, or change all vars..
input <- "cyl"
ggplot(mtcars, aes(mpg, hp)) +
geom_point(aes_string(colour = input))
# or
# ggplot(mtcars, aes_string("mpg", "hp", colour = input)) +
# geom_point()
# you can put the converted string var in the normal aes()
ggplot(mtcars, aes(mpg, hp, colour = !!as.symbol(input))) +
geom_point()
# note: as.name() and as.symbol() are aliases
Created on 2018-11-05 by the reprex package (v0.2.0).