Struggling to setup multi-step SelectInput in Shiny - r

Question

You can see I'm trying to setup three different filters, all of them reactive to each other, but it keeps giving me errors when I test it out.
What I'm trying to accomplish is setting up these filters in a way where say I choose Territory 1, it'll only give me the corresponding states (Ohio) and cities (Columbus and Cleveland). At the same time, say I select both Territory 1 and 4, it will give me the states (Ohio and Michigan) and the cities (Columbus, Cleveland, and Grand Rapids, but not Detroit). At the same time, say I wanted to ignore the territory field and decided to just go right to the State filter, it will give me all the state options (keeping in mind, I haven't chosen anything in territory yet). Same thing if I just wanted to go right to the city filter.
So basically, I want all the filters to be reactively tied to each other, but with no rigid hierarchy where I have to choose territory first, then state, and finally cities.
Was I able to explain that well?
Here is the code.
The setup:
library(shiny)
library(dplyr)
library(highcharter)
df <- structure(list(territory = structure(c(1L, 1L, 2L, 2L, 3L, 4L
), .Label = c("1", "2", "3", "4"), class = "factor"), state = structure(c(3L,
3L, 1L, 1L, 2L, 2L), .Label = c("Indiana", "Michigan", "Ohio"
), class = "factor"), city = structure(c(2L, 1L, 6L, 4L, 3L,
5L), .Label = c("Cleveland", "Columbus", "Detroit", "Gary", "Grand Rapids",
"Indianapolis"), class = "factor"), sales = 5:10, leads = 11:16), class = "data.frame", row.names = c(NA,
-6L)) %>%
mutate_all(as.character)
ui <- {
fluidPage(
fluidRow(
selectizeInput(
inputId = 'selectTerritory',
label = 'Select Territory',
choices = c('All Territories', sort(unique(df$territory))),
multiple = TRUE,
selected = 'All Territories'),
uiOutput(
outputId = 'selectState'),
uiOutput(
outputId = 'selectCity'),
highchartOutput("test")
# plotOutput()
)
)
}
server <- function(input, output, session) {
output$selectState <- renderUI({
# if 'All Territories' is not selected, then filter df by selected Territories. Otherwise, just get all states.
if (!('All Territories' %in% input$selectTerritory)) {
df <- df %>%
filter(
territory %in% input$selectTerritory)
}
states <- sort(unique(df$state))
selectizeInput(
inputId = 'selectState',
label = 'Select State',
choices = c('All States', states),
multiple = TRUE,
selected = 'All States')
})
output$selectCity <- renderUI({
# same strategy
if (!('All States' %in% input$selectState)) {
df <- df %>%
filter(
state %in% input$selectState,
territory %in% input$selectTerritory)
} else {
df <- df %>%
filter(
territory %in% input$selectTerritory)
}
cities <- sort(unique(df$city))
selectizeInput(
inputId = 'selectCity',
label = 'Select City',
choices = c('All Cities', cities),
multiple = TRUE,
selected = 'All Cities')
})
geog <- reactive({
res <- df %>% filter(is.null(input$selectTerritory) | territory %in% input$selectTerritory,
is.null(input$selectState) | state %in% input$selectState,
is.null(input$selectCity) | city %in% input$selectCity)
})
output$test <- renderHighchart({
res <- geog() %>% select_all()
graph <- res %>% group_by_all() %>% summarise(totals=sum(sales))
highchart() %>% hc_add_series(data = graph, type = "bar", hcaes(y = totals),
showInLegend = TRUE) %>% hc_add_theme(hc_theme_flat())
})
}
shinyApp(ui, server)

Answer 1

Firstly, I prefer single-file Shiny apps (it's much easier to copy/paste the whole app rather than having separate files for ui and server).
The other benefit of single-file Shiny apps is that when you post the code to something like Stack Overflow, you can just copy and paste the whole thing, including your calls to library() which are very important! For example, I don't know what package you are using for highChartOutput(), so including those library() calls makes it easier when people are reproducing your code (and becomes the default if you just have a single-file Shiny app). On a related note, your question doesn't really have anything to do with a plot output, so I'm ignoring that part here (you should be able to access the outputs the same way you normally would in a Shiny app).
Secondly, I wouldn't use rbind() to make a df (it can do weird things with the data types). You can just define your variable columns, and directly call data.frame() (e.g., df <- data.frame(v1,v2)). I created the reproducible data by defining your variable columns, calling data.frame, and then running dput(df), which means we can define your dataframe in a single call (and so it's easy to reproduce and avoid any typos).
Thirdly, I'm using dplyr::mutate_all(as.character) because I don't know the package you use for this (taRifx).
Finally, to answer your question... The way I would approach this problem is to define extra variables to each class corresponding effectively to 'All variables', and have these selected as the default for each selectizeInput.
library(shiny)
library(dplyr)
df <- structure(list(territory = structure(c(1L, 1L, 2L, 2L, 3L, 4L
), .Label = c("1", "2", "3", "4"), class = "factor"), state = structure(c(3L,
3L, 1L, 1L, 2L, 2L), .Label = c("Indiana", "Michigan", "Ohio"
), class = "factor"), city = structure(c(2L, 1L, 6L, 4L, 3L,
5L), .Label = c("Cleveland", "Columbus", "Detroit", "Gary", "Grand Rapids",
"Indianapolis"), class = "factor"), sales = 5:10, leads = 11:16), class = "data.frame", row.names = c(NA,
-6L)) %>%
mutate_all(as.character)
ui <- {
fluidPage(
fluidRow(
selectizeInput(
inputId = 'selectTerritory',
label = 'Select Territory',
choices = c('All Territories', sort(unique(df$territory))),
multiple = TRUE,
selected = 'All Territories'),
uiOutput(
outputId = 'selectState'),
uiOutput(
outputId = 'selectCity')
# plotOutput()
)
)
}
server <- function(input, output, session) {
output$selectState <- renderUI({
# if 'All Territories' is not selected, then filter df by selected Territories. Otherwise, just get all states.
if (!('All Territories' %in% input$selectTerritory)) {
df <- df %>%
filter(
territory %in% input$selectTerritory)
}
states <- sort(unique(df$state))
selectizeInput(
inputId = 'selectState',
label = 'Select State',
choices = c('All States', states),
multiple = TRUE,
selected = 'All States')
})
output$selectCity <- renderUI({
# same strategy
if (!('All States' %in% input$selectState)) {
df <- df %>%
filter(
state %in% input$selectState,
territory %in% input$selectTerritory)
} else {
df <- df %>%
filter(
territory %in% input$selectTerritory)
}
cities <- sort(unique(df$city))
selectizeInput(
inputId = 'selectCity',
label = 'Select City',
choices = c('All Cities', cities),
multiple = TRUE,
selected = 'All Cities')
})
}
shinyApp(ui, server)