Predictive modeling success hinges on selecting the features that are most likely to affect the desired outcome and sub-optimal featuring engineering is one of the culprits behind poorly performing models. In this session, we will discuss how a small number of carefully curated features can successfully serve as a proxy of predictive characteristics across a wide range of applications. Learn how data enrichment and the use of pre-defined features can simplify feature engineering, make it more science than art, and eventually improve model quality.
Anindya Datta, CEO - Mobilewalla
Sean Welch, Host and Producer - Data Science Central