A Quick Overview on the Kaggle Competition for Avito

I didn't have much time for this competition, so didn't invest much into feature engineering, creating ensembles or other things. As I participated in the Avazu competition as well, which included the use of tinrtgu's now-famous code, I decided to use the same approach here. The overall goal of the competition is to analyze user behavior in order to generate a model for recommending ads to be shown in front of users, with the success metric being whether or not the user clicks on the ad. There is already a lot of work on this topic, so there is no need to rebuild everything from scratch. If you haven't read the paper from Google on FTRL for ad prediction and their view from the trenches then I can really recommend that as a first step.