shiny app fails on shinyapps.io, data not found

1
votes

I have a that runs fine in locally but not on shinyapps.io. I consulted the Rstudio guidance on this point and none of the obvious problems seem present (e.g. absolute paths). The problem seems to be that the app doesn't find the data; I have checked and rechecked and am uploading the data to shinyapps.io with the code. The error in the log file is:

Warning: Error in is.data.frame: object 'dat' not found.

This is the server.R file:

library(ggplot2)

layerFit <- function(fit){
  geom_line(data = fit,
            aes(x, y), color="red",size = 1, linetype = 2)
}

dat <- read.csv("data.csv", stringsAsFactors = F, na.strings = "NA")

dat$date <- as.Date(paste0(dat$month, "-", dat$day), format = "%m-%d")
dat$time <- 1:nrow(dat)

server = function(input, output) {

  expfit <- reactive({
   dat2 <- dat[!is.na(dat[[input$variable]]),]
    mod <- lm(as.formula(paste("log(",input$variable, ") ~ time")), dat2)
    expfit <- data.frame(x = dat2$date,
                         y = exp(fitted(mod)))

    return(expfit)
  })


  output$data <- renderPlot({
    fit <- expfit()
    p <- ggplot(dat, aes(y = !!input$variable, x= date))+
      geom_point() +geom_line()+
    if(input$toggleExpFit) p = p + layerFit(fit)
    p
  })
}

And this is the ui.R file:

ui <- fluidPage(
  varSelectInput("variable", "Variable:", dat[,c("x","y")]),
  checkboxInput("toggleExpFit", label = "Exponential model", value = FALSE), 
  plotOutput("data")
)

You can create fake data like the data.csv file by:

dat <- data.frame(day  1:20, month = 1, x = rnorm(20), y = rnorm(20))

Thanks for your help.

2
rshinyrstudioshinyapps
Does this answer your question? R Shiny read csv file – A. Suliman
I see some parallels but still not sure how to resolve the issue...am a shiny novice. So I will mark 'no' it does not answer the question but continue to look at this... – tvg
It's the same problem which is caused by calling dat in the ui.R "try to clean the Global env. and rerun, the app will fail". Another solution, put dat <- read.csv("data.csv", stringsAsFactors = F, na.strings = "NA") dat$date <- as.Date(paste0(dat$month, "-", dat$day), format = "%m-%d") dat$time <- 1:nrow(dat) in global.R which is a script in the same folder as server.R and ui.R. you may add layerFit <- function(fit){ geom_line(data = fit, aes(x, y), color="red",size = 1, linetype = 2) } to global.R to organize your code. – A. Suliman
THAT answers the question, thank you. I couldn't see how to solve the problem of calling dat in ui.R from the other question, probably because I am very new at shiny. – tvg
Hi! Shiny app file works the same as a project file in R. Then every file ypu're calling locally should be within the project directory, i.e. under the shiny app file for the app to run in shinyapps.io – Gibran Peniche

2 Answers

1
votes

As suggested by A. Suliman, the solution is to put everything before the server function into a separate global.R script.

1
votes

As described, you could work with a global.R script.

Alternatively, you could largely keep your code as is if you remove the choices from varSelectInput(), e.g. varSelectInput("variable", "Variable:", ""). Then, in your server.R file, you could add an updateVarSelectInput() statement. That way, you could even add interactive filters over your dat variable and the choices would change accordingly. You wouldn't necessarily need a global.R file in that case.

See ?updateSelectInput for a nice example.