
Chapter 5. Can you predict the customers who are likely to leave? 89
The second line in Figure 5-3 has 20% of the data set and it shows the
distribution of the variables who already left the company, in other words,
churners in contrast to the whole population. As you can see, the distribution of
each variable is quite different from the first line. These customers have older
handsets and they use less telephone numbers when they call. Their quality of
call is lower than the general.
Using bivariate statistics, you know that NEW_HANDSET, STATUS,
OUTSPHERE_RANGE, QUALITY, CONTRACT EX are the five most significant
variables to distinguish churn ...