Practical Predictive Analytics

How survSplit works

We will look a bit closer at the survSplit function and how it affects the satisfaction variable. Since we need to keep track of how customer satisfaction changed at time period 6, survSplit function will alter the dataframe by creating new rows and adjusting the time periods to reflect the change in customer satisfaction. In our example, that means that the function would create additional rows after the 6 months cutoff (cut=6) based upon the values of Xtenure2.

As an example of how this would work, first look at the original ChurnStudy dataframe and observe that the very first record has a tenure of 7 months. This customer would have been around long enough for a second survey:

 ChurnStudy$seqid <- seq(1:nrow(ChurnStudy)) ...