Predicting Customer Value Using Hazard and Intensity Models
Duration: 2.0 days
About the Instructor
This course is taught by the M2001 and M2002 co-chair Will Potts. Will has more than 15 years experience as a statistical consultant in science and industry. Currently, he is the Chief Statistician at Data Miners, Inc. where he collaborates on data analysis projects for clients from many industries. Will Potts is the developer of several popular training courses including Survival Data Mining, Neural Network Modeling, and Predictive Modeling using Logistic Regression. Prior to Data Miners, Will was the Co-Director of the Biometrical Consulting Service at the Beltsville Agricultural Research Center, a Senior Biostatistician at the Cleveland Clinic Foundation, and a Statistical Services Specialist at SAS Institute.Audience
The intended audience of this course is predictive modelers who mine company databases using SAS software.Course Description
The future value of a contractual customer depends on the remaining lifetime of their products and services. Predicting customer value involves modeling the churn hazard. The value of a non-contractual customer can be predicted by modeling the hazard of silent churn. Alternatively, the value of a non-contractual customer can be predicted by modeling the intensity of their future purchases.This two-day course covers statistical methods for
- Estimating the mean and median residual life
- Modeling the intensity function of recurrent events
- Forecasting values of time-dependent covariates.
Prerequisites
The course contains a brief overview of predictive hazard modeling. However, we recommend that students first attend the Survival Data Mining: Predictive Hazard Modeling for Customer History Data course.