Mastering Recommender Systems

Explore strategies employed by Kaggle Grandmasters of NVIDIA to secure top positions in a data science competition focused on building a high-functioning recommendation system for e-commerce. Learn about the two-stage model approach, including candidate generation using co-visitation matrices and reranker model development with feature selection and engineering. Discover insights from second and third-place solutions, model ensembling techniques, and participate in a Q&A session. Gain valuable knowledge about recommender systems, data science competitions, and advanced techniques used by industry experts in this 47-minute video from the Grandmaster Series.

– Introduction
– Overview & Summary of the Challenge
– Recommender Systems - 2 Stage model
– Stage 1: Candidate Generation & Co-visitation matrices
– Co-Visitation matrices explained
– Stage 2: Reranker model - Feature selection & engineering
– Second-place solution
– Third-place solution
– Model Ensembling
– Q&A Session