RStudio-Shiny code works line-by-line (Ctrl+Enter), but not with the "Run App" button

Question

in RStudio the below Shiny code works fine if I run it using Ctrl+Enter, line-by-line. However, if I run the whole code using the "Run App" button it generates this error:

Error in ts(x) : 'ts' object must have one or more observations

I think it is due to "lambda" parameter but I cannot see why. Any help is appreciated.

The link for "data.csv" is https://www.dropbox.com/s/p1bhacdg8j1qx42/data.csv?dl=0

====================================

library(shiny)
library(shinydashboard)
library(plotly)
library(forecast)

df <- read.csv("data.csv")
demand <- ts(df$demand, start = c(1995, 1), frequency = 12)

lbd <- BoxCox.lambda(demand, lower=-5, upper=5)
m <- ar(BoxCox(demand,lambda=lbd))
fit_BC <- forecast(m, h=12, lambda=lbd)

ui <- dashboardPage(
  dashboardHeader(title = "Plot"),
  dashboardSidebar(disable = TRUE),
  dashboardBody(fluidRow(column(width = 12, box(plotlyOutput("forecast_plots"),width = NULL))))
)

server <- function(input, output) {
  output$forecast_plots <- renderPlotly({
    autoplot(fit_BC)
  })
}

shinyApp(ui, server)

==================================

Answer 1

autoplot() returns ggplot object. But your output$forecast_plots requires plotly object(with plotlyOutput() function).

Working code is like the following:

ui <- dashboardPage(
    dashboardHeader(title = "Plot"),
    dashboardSidebar(disable = TRUE),
    dashboardBody(fluidRow(column(width = 12, box(plotOutput("forecast_plots"),width = NULL))))
)

server <- function(input, output) {
    output$forecast_plots <- renderPlot({
        autoplot(fit_BC)
    })
}

ggplot objects can be easily converted with ggplotly function, but unfortunately converted plotly autoplot graph loses the forecasting region. You can verify it like:

ui <- dashboardPage(
    dashboardHeader(title = "Plot"),
    dashboardSidebar(disable = TRUE),
    dashboardBody(fluidRow(column(width = 12, box(plotlyOutput("forecast_plots"),width = NULL))))
)

server <- function(input, output) {
    output$forecast_plots <- renderPlotly({
        ggplotly(autoplot(fit_BC))
    })
}

Add

I found autoplotly library.https://terrytangyuan.github.io/2018/02/12/autoplotly-intro/

autoplotly() function can convert autoplot object to plotly object which is roughly correct.

library(shiny)
library(shinydashboard)
library(plotly)
library(forecast)
library(autoplotly)

df <- read.csv("c:/Users/010170283/Downloads/data.csv")
demand <- ts(df$demand, start = c(1995, 1), frequency = 12)

lbd <- BoxCox.lambda(demand, lower=-5, upper=5)
m <- ar(BoxCox(demand,lambda=lbd))
fit_BC <- forecast(m, h=12, lambda=lbd)

ui <- dashboardPage(
    dashboardHeader(title = "Plot"),
    dashboardSidebar(disable = TRUE),
    dashboardBody(fluidRow(column(width = 12, box(plotlyOutput("forecast_plots"),width = NULL))))
)

server <- function(input, output) {
    output$forecast_plots <- renderPlotly({
        autoplotly(autoplot(fit_BC))
    })
}

shinyApp(ui, server)

The forecast region can be seen with it, and hi/lo 80 % edge values are presented with mouse hover event.