Although we've already looked at dygraphs, it's worth looking at it again, so we can see how to build a prediction in the final plot. This is quite simple to do, and the particular prediction statistics that we will use here has few assumptions about the data and can be used in most contexts. Before we take a look at the code, let's take a look at the final application:

As you can see, the graph contains the actual data as well as a prediction of how the data might look over the next few years. The blue shading indicates prediction intervals, which give us an idea of the reliability of the projection. Let's now turn our attention to the code to produce this plot:

output$predictSeries <- renderDygraph({

Again, the graph is produced using the special renderDygraph() function.

  theSeries <- group_by(passData(), yearmon) %>%
  summarise(meanSession = mean(sessionDuration, na.rm = TRUE),
    users = sum(users), sessions = sum(sessions)
  )

As mentioned earlier...