The Power of Personalized Customer Experiences Through AI Innovations
summary
Describes a practical build of a recommender for a sports retailer: data aggregated with pandas, features encoded with count vectorizer, similarity scored via cosine similarity, and demoed to match three new promos to top three customers each. Product stack uses Spark preprocessing for terabyte-scale pipelines and offers a dozen recommenders (bestseller/trending fallbacks, content‑based, SVD collaborative filtering, GNN). Reported client uplift moves conversion rates from about 10% up to 35%.
AI-written summary · about
chapters (7)
- 0:00Welcome
- 1:25Personalization importance
- 3:52Think Solver overview
- 8:02Think Solver personalization
- 12:22Recommender challenges
- 20:05Recommender engine demo
- 27:13Q&A
related talks
29mThe Challenges of Building a Personalized Recommendation System in E-Commerce
32mLeveraging recommender systems to personalise search results
41mRecommender Systems: Crafting Effective Personalised Experiences
27mTransform online shopping experiences with personalized recommendations
38mRecommendations at Rappi: from MVP to Data Product
26mLarge-Scale Recommendation System with Python and Spark
27mEnhancing Personalization Using Recommendation Systems & Multi-Armed Bandits
31mCreating recommender systems with Spark, Scala and Prediction.Io