Reactive Data Issues in RShiny

Question

I'm working on a Shiny dashboard for a personal project with some football stats. Whenever I change the statistic to be graphed and/or the filter, I get the same players that were in the first dataset. For example, when I start the app, the app creates a graph of the top ten rushers in school history with a filter of rushing attempts >= 0. When I change the statistic selection to rushing average, however, those ten players are the ones shown, which is incorrect.

library(readxl)
library(tidyverse)
library(purrr)
library(shiny)

interface <- fluidPage(
    titlePanel(" "), 
    sidebarLayout(
        sidebarPanel(
            h1("Stats!"), 
            selectInput("stat_selection", 
                        label = "Select a season statistics", 
                        choices = c("Rushing Yards",
                                    "Rushing Touchdowns",
                                    "Rushing Average",
                                    "Reciving Yards",
                                    "Receptions",
                                    "Receiving Touchdowns",
                                    "Receiving Average"),
                        selected = "Rushing Yards"), 
            selectInput("filter_input", 
                        label = "Select a statistic to filter by", 
                        choices = c("Rushing Yards",
                                    "Rushing Touchdowns",
                                    "Rushing Average",
                                    "Rushing Attempts",
                                    "Reciving Yards",
                                    "Receptions",
                                    "Receiving Touchdowns",
                                    "Receiving Average"),
                        selected = "Rushing Attempts"), 
            numericInput("filter_number", 
                         label = "Type a number for the filter (>=)", 
                         value = 0, min = 0), 
            actionButton("button", "Graph")), 
        mainPanel(
            plotOutput("plot_button"),
            tableOutput("table_button")
        )
    )
)

server_osu <- function(input, output) {
    dataInput <- reactive({
        switch(input$stat_selection, 
               "Rushing Yards" = rush_yds, 
               "Rushing Touchdowns" = rush_tds, 
               "Rushing Average" = rush_avg,
               "Reciving Yards" = rec_yds,
               "Receptions" = rec_rec,
               "Receiving Touchdowns" = rec_td,
               "Receiving Average" = rec_avg)
    })
    filterInput <- reactive({
        switch(input$filter_input, 
               "Rushing Yards" = rush_yds, 
               "Rushing Touchdowns" = rush_tds, 
               "Rushing Average" = rush_avg,
               "Rushing Attempts" = rush_att,
               "Reciving Yards" = rec_yds,
               "Receptions" = rec_rec,
               "Receiving Touchdowns" = rec_td,
               "Receiving Average" = rec_avg)
    })
    filter_number <- reactive(as.double(input$filter_number))
    table_button_react <- eventReactive(input$button, {
        dataset <- dataInput()
        val <- filter_number()
        colnames(dataset)[1] = "Player and Season"
        dataset_filter <- filterInput()
        colnames(dataset_filter)[1] = "Player and Season"
        dataset <- left_join(dataset, dataset_filter)
        colnames(dataset)[1] = "Player and Season"
        og <- colnames(dataset)[3]
        colnames(dataset)[3] = "filter"
        original <- colnames(dataset)[2]
        colnames(dataset)[2] = 'selected'
        dataset <- dataset %>% 
            filter(filter >= val)
        dataset <- dataset %>% 
            top_n(10) %>% 
            arrange(-selected)
        colnames(dataset)[2] = original
        colnames(dataset)[3] = og
        dataset
    })
    plot_button_react <- eventReactive(input$button, {
        dataset <- dataInput()
        val <- filter_number()
        colnames(dataset)[1] = "Player and Season"
        dataset_filter <- filterInput()
        colnames(dataset_filter)[1] = "Player and Season"
        dataset <- left_join(dataset, dataset_filter)
        colnames(dataset)[1] = "Player and Season"
        colnames(dataset)[2] = "selected"
        colnames(dataset)[3] = "filter"
        dataset <- dataset %>% 
            filter(filter >= val)
        top_ten <- dataset %>% top_n(10)
        min = min(top_ten$selected)
        max = max(top_ten$selected)
        ggplot(top_ten, aes(x = reorder(`Player and Season`, -selected), y = selected)) +
            geom_bar(stat = 'identity') + theme_minimal() + xlab('SEASON') +
            ylab(input$stat_selection) + theme(text=element_text(size=16)) +
            scale_fill_manual(values = c('#BBBBBB', '#BB0000')) +
            theme(axis.text.x = element_text(angle = 45, hjust = 1)) +
            theme(legend.position = 'none') +
            coord_cartesian(ylim=c(min - 0.05*min, max + 0.05*max)) +
            theme(axis.title.y = element_text(margin = margin(t = 0, r = 10, b = 0, l = 10))) +
            theme(axis.title.x = element_text(margin = margin(t = 10, r = 0, b = 10, l = 0))) +
            theme(axis.text.y = element_text(size=14),
                  axis.title=element_text(size=16,face='bold')) + 
            labs(caption = '')
    })
    output$plot_button <- renderPlot({
        plot_button_react()
    })
    output$table_button <- renderTable({
        table_button_react()
    })
}

Answer 1

As I mentioned above, a reprex - including input data - would help us to help you. That said, I think the problem is that your XXX_button_react s depend only on input$button . They don't depend on input$stat_selection , input$filter_number or input$filter_input . That's why they don't update as you want them to.

The fix is easy. Just add them (in a call to req() if you like) at the top of each XXXX_button_react , for example:

plot_button_react <- eventReactive(input$button, {
  input$stat_selection
  input$filter_number
  input$filter_input

  <your code here>
})

As a point of style, I feel it's better to separate data generation from data presentation . It makes the logic of your code more obvious, reduces the chance of errors, reduces the need for code duplication and makes your code more reusable.

In your case, I would create a reactive that holds the data you wish to tablulate and plot and then reference that reactive in each of your render_XXXX functions. That would also remove the need for your input$button : the plot and graph would each update automatically whenever you changed one of your other input widgets.