Abstract (Maximum of 2,000 Characters):  Briefly describes content of case.

VMW is a leader in software virtualization with ~6.5 BB annual revenue. VMW sells Workstation that can be bought online (store.vmware.com) and is used for running Mac on Windows. Workstation forms a significant portion of store revenues and most of it is bought online. There is rich digital/clickstream data for the visitors which can be combined with their past purchase history and other offline features as well. The business would like to increase sales of the product by targeting the right customers and needs a propensity model to be built using machine learning that can target the right set of customers.

Michael Butler, the WW head of the store wants to leverage Parag's data sciences team to help him target the right workstation prospects that visit the store. A business conversation between Michael and Parag is followed by a technical discussion between Ravi, the data scientist and Parag. The following are the key questions that Ravi seeks to answer:

• Cross-validation and evaluation in the context of huge imbalance in the data
• Feature selection techniques
• Communicating internal results such as lift curves back to the business
• Different modeling approaches that can be followed
• Interpreting the results for business decision making

Learning Objective (Maximum of 500 Characters):  Briefly describes teaching goals of case.

This case may be used in a Business Analytics & Analytics course or Data Science in Business course of MBA or executive MBA as well as in a PhD/Fellow Program as an elective. The learning objectives of this case are as follows.

• Translate the business problem into a analytics problem
• Learn to apply the right cross-validation in different scenarios online
• Choosing the right evaluation metric in an imbalanced dataset
• Demonstrate application of penalized logistic regression, random forest, and boosting on the imbalanced datasets
• Conceptual understanding of difference between logistic regression and penalized logistic regression, bagging and boosting
• Demonstrate performance of different modeling techniques on the datasets

Come up with use-cases of how the propensity model is used in the e-commerce world