Qini-based uplift regression

Uplift models provide a solution to the problem of isolating marketing effect campaign. For customer churn reduction, uplift are used identify customers who likely respond positively retention activity, only if targeted, and avoid wasting resources on that very switch another company. In practice, performance is measured by Qini coefficient. We introduce Qini-based regression model analyze large insurance company’s Our approach based logistic models. show Qini-optimized acts as regularizing factor for uplift, much penalized likelihood does regression. This results in interpretable with few relevant explanatory variables. parameter estimation significantly improves prediction