Shiny: calculate cumsum based on dygraphs' RangeSelector

Question

I\'m building a shiny app where I want to plot a dataset with one of the variables being a cumulative sum of another variable. The latter needs to be re-calculated every time th

Answer 1

This should do the job, I think it is cleaner to have separate outputs for your dashboard

library(xts)
library(shiny)
library(shinydashboard)
library(dygraphs)

ui <- dashboardPage(
  dashboardHeader(),
  dashboardSidebar(),
  dashboardBody(
    dygraphOutput('plot1'),
    textOutput("cumsum1")
  )
)

server <- function(input, output, session) {

  m_df <- data.frame(date=as.Date(zoo::as.yearmon(time(mdeaths))), Y=as.matrix(mdeaths))

  output$plot1 <- renderDygraph({
    input_xts <- xts(select(m_df, -date), order.by = m_df$date)

    dygraph(input_xts) %>% 
      dyRangeSelector()
  })

  output$cumsum1 <- renderText({
    req(input$plot1_date_window)
    subdata <- cumsum(m_df$Y[m_df$date >= as.Date(input$plot1_date_window[1]) & m_df$date <= as.Date(input$plot1_date_window[2])])
    subdata
  })

}

shinyApp(ui, server)

Answer 2

The problem with your updated code is, that you didn't keep the date information. Also once you start rendering a plot based on a change of the plot itself (recursion) it gets a little tricky. You have to make sure that re-rendering the plot doesn't trigger the rendering again or you'll end up in a loop. That's why I set retainDateWindow = TRUE. Besides that you don't want the plot to re-render right away after the first change of the slider that's why I debounced the subdata.

Nevertheless, using dygraphs you still have the problem, that when you add cumsum as a series your plot for dyRangeSelector is changed (y maximum of all series). Please see the following code:

library(xts)
library(shiny)
library(shinydashboard)
library(dygraphs)
library(dplyr)

ui <- dashboardPage(
  dashboardHeader(),
  dashboardSidebar(),
  dashboardBody(
    dygraphOutput('plot1')
  )
)

server <- function(input, output, session) {

  m_df <- data.frame(date=as.Date(zoo::as.yearmon(time(mdeaths))), Y=as.matrix(mdeaths))

  subdata <- reactive({
    if(!is.null(input$plot1_date_window)){
      subdata <- m_df[m_df$date >= as.Date(input$plot1_date_window[1]) & m_df$date <= as.Date(input$plot1_date_window[2]), ]
      subdata$cumsum <- cumsum(subdata$Y)
      subdata$Y <- NULL
    } else {
      subdata <- NULL
    }

    return(subdata)
  })

  subdata_d <- subdata %>% debounce(100)

  output$plot1 <- renderDygraph({
    input_xts <- xts(select(m_df, -date), order.by = m_df$date)
    if(is.null(subdata_d())){
      final_xts <- input_xts
    } else {

      subdata_xts <- xts(select(subdata_d(), - date), order.by = subdata_d()$date)
      final_xts <- cbind(input_xts, subdata_xts)
    }

    p <- dygraph(final_xts) %>% dySeries(name="Y") %>%
      dyRangeSelector(retainDateWindow = TRUE)

    if("cumsum" %in% names(final_xts)){
      p <- dySeries(p, name="cumsum", axis = "y2")
    }

    p
  })

}

shinyApp(ui, server)

Just as @PorkChop mentioned I'd recommend multiple outputs for this scenario. Furthermore, I'd suggest to have a look at library(plotly) and it's event_data().