Practical Data Science - Marketing Science

  Discrete Choice Models can be useful to (1) provide insights for user behaviors and (2) enable counter-factual simulations for marketing and production decision support. I will provide some details on the modeling approach, starting with static models.   

A single A/B test metric is rarely enough to have a robust picture of the A/B test. Summarized below are A/B testing Metric Taxonomy, which will be helpful for you to understand what categories of metrics you need to have. 

This 10-page introduction covers the following topics.  Many links to online books, useful blogs, and open-source packages are included.  Prediction vs. Causal Inference Two Traditions and Resource List What Can Go Wrong with Observational Data? Two Approaches in Econometrics: Reduced-form vs. Structural Data Types: cross-sectional data, time-series data, panel data Causal Inference Introduction