15.820 Customer Analytics Using Probability Models
Provides powerful, cutting-edge quantitative tools to apply to problems of genuine managerial interest. Topics include modeling unobserved heterogeneity of customer types, estimating customer lifetime value using "buy until dead" models, understanding market concentration and "80-20 rules," forecasting media exposures, predicting adoption of new products, assessing effectiveness of promotional activities, exploiting demographic information, and measuring brand performance. Mathematical tools include stochastic processes, maximum likelihood estimation, empirical Bayes methods, continuous mixture and latent class modeling, and proportional hazard regression. Students practice how to derive a wide range of probability models from basic principles, estimate the parameters of models on real data sets (using the Solver tool in Excel), and communicate findings to stakeholders. As a final project, each student applies the tools presented to a data set of his or her own choice.
15.820 will not be offered this semester. It will be instructed by M. Braun.
This class counts for a total of 9 credits. This is a graduate-level class.
© Copyright 2015 Yasyf Mohamedali