Practical Predictive Analytics

Incorporating interaction terms

To incorporate these interactions in a future model we specify them directly in the right side of the model formula. For example, this specification includes service calls, as well as the interaction between service calls and the number of purchases last month, as model terms:

CoxModel.2 <- coxph(Surv(time0,Xtenure2, Churn2) ~                      Xeducation + Xgender + Xsatisfaction2 + Xservice.calls + Xpurch.last.month:Xservice.calls +                      Xpurch.last.month + Xmonthly.charges,                    data=SURV2)