The Challenges of Building a Personalized Recommendation System in E-Commerce

on YouTube ↗29m

" summary

Built a personalized recommender for a Serbian e-commerce site from scratch using both content-based filtering and collaborative filtering with ALS matrix factorization, trained on about a year of logs (tens of millions of events) after an initial six months of sparse data. Pipeline runs on PySpark on Hadoop with Zeppelin, model training can take up to a day, evaluation via A/B testing shows higher click and conversion rates, and rollout to all users is imminent.

AI-written summary · about

" chapters (11)

1. 0:00Introduction
2. 0:40Background
3. 2:06Why do we need recommendation systems
4. 3:38Examples
5. 4:21Netflix
6. 6:47ContentBased Filtering
7. 11:01Collaborative Filtering Example
8. 13:23Cold Start Problem
9. 16:16Ad Testing
10. 17:49Personalized Newsletter
11. 19:33Questions

" related talks

• 28mThe Power of Personalized Customer Experiences Through AI InnovationsN. Trajkov · DSC Europe 23
• 32mLeveraging recommender systems to personalise search resultsSoraya Hausl · PyData London 2017
• 26mLarge-Scale Recommendation System with Python and SparkPhil Anderson · PyOhio 2018
• 41mRecommender Systems: Crafting Effective Personalised ExperiencesKim Falk · Øredev 2025
• 37mBuild a Recommender System in Apache Spark and Integrate it Using AkkaWillem Meints · Scala Days Berlin 2016
• 58mRecommender Systems – A WalkthroughArtus Krohn-Grimberghe · PASS Business Analytics Conference 2013
• 48mBuild a Recommender System in Apache Spark and Integrate It Using AkkaWillem Meints · Scala Days Berlin 2016
• 55mBuilding a Recommendation Engine with Machine Learning TechniquesBrian Sam-Bodden · Full Stack Fest 2016