American and British Time Styles – application of K-Means Clustering for market segmentation based on time-use, and demographic profiling of the identified segments.
Predicting the prices for pre-owned Audis – students are introduced to the concept of Hedonic Price Regression. They are asked to develop a linear regression model that best explains the pricing of pre-owned Audis in online classified ads. Then, they are asked to predict the posted prices for a holdout set of re-owned Audis. The grade on this case is proportional to the accuracy (RMSE) of these price predictions.
Diabetes Among Pima Indian Women – students are introduced to response modeling and propensity scoring by developing their own estimator for Logistic Regression on Excel, starting with results from a standard linear regression.
Default on credit-card loans – students first develop their own estimator for Logistic Regression to learn how response modeling works and how to assess the fit of binary response models. They then use KATE to make optimal (maximum cost savings) targeting decisions on a holdout sample and are graded in proportion to the actual cost savings known only to the instructor.
Why are we losing so much human capital? – case focused on the interpretation of Linear and Logistic Regression results (fit, predictive fit and parameter estimates) produced by KATE.
Caravan Insurance – students are asked to develop a propensity-scoring model on a calibration sample and make targeting decisions on a test sample for which they only see the potential predictors. This case including a scoring sheet that computes the net customer lifetime value for each prospect in the test sample.
AFR Lapsed Donors – this is a more advanced case on response modeling than the previous ones. Here, students are asked to make targeting decisions based not only on who is likely to make a donation, but also on how much they are expected to donate.