I would like to extract the element name, and not the specific value, from a list used for the choices argument in selectInput() from R Shiny.
The selectInput function looks like this:
# ...
selectInput("xvar", "What is the predictor variable?",
choices = list("MPG" = "mpg",
"Cylinders" = "cyl",
"Engine Displacement" = "disp",
"Horse Power" = "hp",
"Gears" = "gear"),
# ...
In my server.R code I would like to use, for example, "Cylinders" and not "cyl" as an axis label. For example (using ggplot2):
# ...
labs(x = input$xvar, y = input$yvar) +
# ...
names(input$xvar) returns NULL. Is there any way to call input$xvar and return the name?
Thanks to Paul's comments, the links he provided, and this SO thread, I was able to answer my question.
Below I provide the old ui.R and server.R scripts which generated axis labels I was not happy with, as well as new ui.R and server.R scripts where the axis labels are improved. (Changes in the new scripts are marked with # diff)
old ui.R:
shinyUI(fluidPage(
titlePanel("Fit Regression Line for Chosen Variables and Points"),
sidebarLayout(
sidebarPanel(
h2("Model Specifics"), br(),
selectInput("xvar", "What is the predictor variable?",
choices = list("MPG" = "mpg",
"Cylinders" = "cyl",
"Engine Displacement" = "disp",
"Horse Power" = "hp",
"Gears" = "gear"),
multiple = FALSE),
selectInput("yvar", "What is the outcome variable?",
choices = list("MPG" = "mpg",
"Cylinders" = "cyl",
"Engine Displacement" = "disp",
"Horse Power" = "hp",
"Gears" = "gear"),
multiple = FALSE, selected = "cyl"),
h4("Intercept"), textOutput("int"),
h4("Slope"), textOutput("slope")
),
mainPanel(
br(), h2("Display"), h4("Drag to select which points to include in model"),
plotOutput("plot", brush = brushOpts(id = "brush1"))
)
)
))
old server.R:
shinyServer(function(input, output) {
model <- reactive({
points <- brushedPoints(mtcars, brush = input$brush1,
xvar = input$xvar,
yvar = input$yvar)
if(nrow(points) <= 1) {
return(NULL)
} else {
lm(as.formula(paste0(input$yvar,
"~", input$xvar)),
data = points)
}
})
output$int <- renderText({
if(is.null(model())) {
"Too few data points selected"
} else {
round(model()[[1]][1], 2)
}
})
output$slope <- renderText({
if(is.null(model())) {
"Too few data points selected"
} else {
round(model()[[1]][2], 2)
}
})
output$plot <- renderPlot({
library(ggplot2)
ggplot(mapping = aes(x = mtcars[, input$xvar],
y = mtcars[, input$yvar])) +
theme_minimal() +
geom_point() +
labs(x = input$xvar, y = input$yvar) +
coord_cartesian(x = c(0, 1.2*max(mtcars[, input$xvar])),
y = c(0, 1.2*max(mtcars[, input$yvar]))) +
if(!is.null(model())) {
geom_abline(intercept = model()[[1]][1], slope = model()[[1]][2],
colour = "red", lwd = 2, alpha = 0.3)
}
})
})
Changes in the scripts are marked with # diff
new ui.R:
shinyUI(fluidPage(
titlePanel("Fit Regression Line for Chosen Variables and Points"),
sidebarLayout(
sidebarPanel(
h2("Model Specifics"), br(),
uiOutput("si_xvar"), # diff
uiOutput("si_yvar"), # diff
h4("Intercept"), textOutput("int"),
h4("Slope"), textOutput("slope")
),
mainPanel(
br(), h2("Display"), h4("Drag to select which points to include in model"),
plotOutput("plot", brush = brushOpts(id = "brush1"))
)
)
))
new server.R:
shinyServer(function(input, output) {
varlist <- list("MPG" = "mpg", # diff
"Cylinders" = "cyl",
"Engine Displacement" = "disp",
"Horse Power" = "hp",
"Gears" = "gear")
output$si_xvar <- renderUI( # diff
selectInput("xvar", "What is the predictor variable?",
choices = varlist,
multiple = FALSE)
)
output$si_yvar <- renderUI( # diff
selectInput("yvar", "What is the outcome variable?",
choices = varlist,
multiple = FALSE, selected = "cyl")
)
model <- reactive({
points <- brushedPoints(mtcars, brush = input$brush1,
xvar = input$xvar,
yvar = input$yvar)
if(nrow(points) <= 1) {
return(NULL)
} else {
lm(as.formula(paste0(input$yvar,
"~", input$xvar)),
data = points)
}
})
output$int <- renderText({
if(is.null(model())) {
"Too few data points selected"
} else {
round(model()[[1]][1], 2)
}
})
output$slope <- renderText({
if(is.null(model())) {
"Too few data points selected"
} else {
round(model()[[1]][2], 2)
}
})
output$plot <- renderPlot({
library(ggplot2)
ggplot(mapping = aes(x = mtcars[, input$xvar],
y = mtcars[, input$yvar])) +
theme_minimal() +
geom_point() +
labs(x = names(which(input$xvar == varlist)), # diff
y = names(which(input$yvar == varlist))) + # diff
coord_cartesian(x = c(0, 1.2*max(mtcars[, input$xvar])),
y = c(0, 1.2*max(mtcars[, input$yvar]))) +
if(!is.null(model())) {
geom_abline(intercept = model()[[1]][1], slope = model()[[1]][2],
colour = "red", lwd = 2, alpha = 0.3)
}
})
})