noritanushree / gapminder-app

Shiny application for viewing Gapminder data

Gapminder-app

Shiny application for viewing Gapminder data.

• View Gapminder Shiny app hosted by shinyapps.io
• View code: server.R and ui.R

Reflections

Debugging Shiny app issues was not easy, and very time consuming!

• If anything, making my first Shiny app helped me to appreciate the awesomeness of built in testing when building an R package.

Adding a radioButtons widget to Shiny app, and using selection from radioButtons to filter (using dplyr) and plot (using ggplot) Gapminder data was MUCH harder than I expected!

• This step was very painful and time consuming to debug. The choices within radioButtons are returned in quotes, and by default, dplyr and ggplot do not handle quoted variables. Thus the output of renderTable and renderPlot failed.
• I tried many different ways to remove quotes from the variables, but this turned out to be difficult.
• Solution part 1: I found this StackOverflow issue related to my problem. When I added matches(input$variable) to the reactive function, the downstream renderTable function works, but renderPlot didn't work.
• Solution part 2: I also needed to use aes_string within the renderPlot ggplot call! Thanks to both Winston's post on Shiny google groups and Julia's help (see issue 1)

The reactive function is very confusing!

• From the earliest stages of the Shiny app development, I obtained the error: "Error in eval(expr, envir, enclos) : object 'year' not found". I spent hours trying to debug this issue. It turns out I had to add another is.null argument for the year variable produced by the reactive function, but not in the way I expected...
• Clue: adding the print statement str(two_country_data$year()) produced the following error: "Error in two_country_data$year : object of type 'closure' is not subsettable".
• i.e. reactive objects are functions!!! Consequently, if you produce a dataframe using a reactive function, you do not access columns in the usual way. To access reactive object variable columns, I didn't realise you had to add open close brackets BEFORE the $ column name! i.e. two_country_data()$year (correct) is not the same as two_country_data$year() (incorrect).
• Thanks again to Julia for helping me figure this out! (see issue 1)

Favourite moment during Shiny app development: no more red error messages popping up on the screen

• I could then concentrate on making the user interface pretty :)

Future applications for Shiny?

• Shiny + ggvis, full integration? It would be nice to hover your mouse over graphs in Shiny and see the values.
• Aim to be as cool as the Google Public Data Explorer (which actually acquired Gapminder's Trendalyzer software!). The play button feature is pretty wicked.
• Add Shiny apps to presentations, e.g. add it as a slide in Keynote/Powerpoint/other? That would be sweet!