Introduction to the OTTO competition on Kaggle (RecSys)

In this video, we meet several very powerful RecSys concepts that can be extended to real-life scenarios!

0:00 Competition intro
1:09 Data overview
6:40 Baseline
8:54 Co-visitation matrix
12:53 Candidate reranking using static rules
20:25 Second-stage Ranker
22:15 Word2vec
28:39 Matrix Factorization
32:05 What's next

The presented concepts build on one another. Furthermore, the OTTO dataset is a very interesting and elegant one, as many features can be created from essentially 4 columns!

And here are the notebooks mentioned in the video:

[ChaiTimeDataScience]

This is legendary and just in time for the last 3 weeks of the competition 🙏 I hope you keep doing this for more competitions

[深度碎片]

I watched half way through and I just want to say that it's amazing how many new insights and ideas (todos) can come out from the notebooks (which I thought I knew) when listening to a great guide (Radek) talking about them

[adityabhatt]

Thank you so much. It will be great if you can make videos on other competitions as well. Will be helpful for beginners.

[vijayendrasdm]

Hi Radek
Thanks for amazing intro to otto(recsys). I have a question on co-visitation matrix

Given aid, you are looking at other aids in that session to build co visit matrix . Now There is a filter you define for session at 9.35 secs which is 1 day .
Does that mean, there are sessions are more than a day long ?