create empty dygraph in Rshiny - r

Question

I'm using Dygraphs package for visualizing actual and the time series predicted values in R shiny. Here is the sample code that I used to generate the Dygraph. In some cases where the data points are less Holt Winters(gamma =T) does not give any prediction and I need to show an empty Dygraph with the title "Insufficient Data"). I'm not able to do this. Appreciate any help on this
library(dygraphs)
plotDyg <- fluidPage(
fluidRow(
box(selectizeInput("c1", "Enter a key",
choices = reactive({sort(unique(df$key))})(),
multiple = FALSE),width=3),
box(dygraphOutput("tsDy"), width = 10, height = 500))
)
ui <- dashboardPage(
dashboardHeader(title = "XYZ"),
dashboardSidebar(
sidebarMenu(
menuItem("abc", tabName = "sidebar2", icon = icon("bar-chart") ,
menuSubItem("def",icon = icon("folder-open"), tabName = "subMenu1")
)
)
),
dashboardBody(
tabItems(
tabItem(tabName = "subMenu1",
fluidRow(
tabBox(
title = "ghi", id = "tabset2",height = "1500px",width = 100,
tabPanel("abcdef", plotDyg)
)
)
)
)
)
)
server <- function(input, output) {
output$tsDy <- renderDygraph({
if(!is.null(input$c1)){
df.0 <- reactive({df[df$key == input$c1,]})()
tspred <- reactive({
df.0 <- convert_to_ts(df.0) # converts column "fin_var" to a monthly time series and returns the entire dataframe
act <- df.0$fin_var
hw <- tryCatch(HoltWinters(df.0$fin_var), error=function(e)NA)
if(length(hw) > 1){
p <- predict(hw, n.ahead = 12, prediction.interval = TRUE, level = 0.95)
all1 <- cbind(act, p)
}else{all1 <- matrix()}
})
if(!is.na(tspred())){
dygraph(tspred(), main = "TS Predictions") %>%
dySeries("act", label = "Actual") %>%
dySeries(c("p.lwr", "p.fit", "p.upr"), label = "Predicted") %>%
dyOptions(drawGrid = F) %>%
dyRangeSelector()
}else{dygraph(matrix(0), main = "Insufficient Data")} # I could just do 'return()' but I want to show an empty Dygraph with the title
}else{return()}
})
}

Answer 1

I too am unable to render Dygraphs with an empty time series. To render a message to the user I used the validate/need functions in Shiny
In your case I would replace
if(!is.na(tspred())){
With
validate(need(!is.na(tspred())), "Insufficient Data"))
This will avoid the "error: argument is of length zero" message within Dygraphs and print an appropriate message to the end user.