Mastering Recommender Systems | Grandmaster Series E8

Video Description

In this episode, we discuss the strategies employed by the participating members of the Kaggle Grandmasters of NVIDIA to emerge as the top three finalists in a recent data science competition. In episode 8 of the Grandmaster Series, delve into the intriguing world of Recommender Systems. In this episode, we discuss the strategies employed by the participating members of the Kaggle Grandmasters of NVIDIA to emerge as the top three finalists in a recent data science competition. The challenge was to construct a high-functioning recommendation system for an e-commerce use case. It required the participants to leverage a dataset that included millions of anonymized user interactions and history to build an effective recommendation strategy. Join us to discover more about their approach! Don't forget to subscribe to our YouTube channel for monthly episodes of the Grandmaster Series. Video Chapters: 00:00 – Introduction 04:27 – Overview & Summary of the Challenge 06:12 – Recommender Systems - 2 Stage model 07:30 – Stage 1: Candidate Generation & Co-visitation matrices 08:54 – Co-Visitation matrices explained 14:56 – Stage 2: Reranker model - Feature selection & engineering 17:21 – Second-place solution 25:51 – Third-place solution 40:23 – Model Ensembling 40:58 – Q&A Session Additional Resources: Kaggle Competition Page: https://www.kaggle.com/competitions/otto-recommender-system/ Learn how to accelerate Pandas DataFrames: https://courses.nvidia.com/courses/course-v1:DLI+T-DS-01+V1/ Learn more about RAPIDS: https://developer.nvidia.com/rapids NVIDIA Accelerated Data Science: https://www.nvidia.com/en-us/deep-learning-ai/solutions/data-science/ About our presenters: Host Jay Rodge is an AI/ML Product Marketing Manager at NVIDIA, driving launches and product marketing initiatives for data science. He holds a master's degree in computer science from Illinois Tech. Benedikt Schifferer is a Deep Learning Engineer at NVIDIA and has been working on recommender systems. He holds a master of science in data science from Columbia University and previously developed recommender systems for a German e-commerce company. Chris Deotte is a Senior Data Scientist at NVIDIA. Chris has a Ph.D. in mathematics and computational science with a thesis on optimizing parallel processing. Chris is a 4x Kaggle grandmaster. Kazuki Onodera is a Senior Data Scientist at NVIDIA. He's been a Kaggle competition Grandmaster since 2019 and he's had six top-two competition rankings Gilberto Titericz, known as Giba is currently a Senior Data Scientist at NVIDIA. Prior to NVIDIA, he worked at Airbnb, Petrobras, and Siemens. Gilberto had held the #1 position at Kaggle for more than two years. Theo Viel is a senior Deep Learning Data Scientist at NVIDIA. Theo holds a master's in Computer Science and recently joined NVIDIA late last year. He’s also a 3x Kaggle Grandmaster. #Kaggle #NVIDIA #datascience Kaggle, Data Science, RAPIDS, NVIDIA, Kaggle Competition, Recommender Systems, RecSys, OTTO competiton