Variable selection for uplift modeling

Insurance companies heavily fund marketing campaigns such as, for instance, customer retention or cross-sell initiatives. Uplift modeling aims at predicting the causal effect of an action such as medical treatment or a marketing campaign on a particular individual by taking into consideration the response to an action. Typically, the result of an uplift model is used to call customers for marketing some products based on important attributes of a customer. The general objective of the research project is to develop an automatic method that will enable practitioners to save time and to follow precise guidelines in order to choose attributes of a customer.

Faculty Supervisor:

Alejandro Murua

Student:

Mouloud Belbahri

Partner:

TD Insurance

Discipline:

Mathematics

Sector:

Information and communications technologies

University:

Program: