Kaggle overview

Kaggle contests are based on the famous Netflix data prediction challenge. After spending as much time as possible with their movie prediction algorithm, they decided to publish training data and put up million dollar prize!

Turns out this was money well spent. This is a supervised learning problem where the data is very sparse. Teams formed, combining best practices and finally were able to reach the prize.

This an an example of a pure prediction problem. All that matters is performance. Another factor, interpretability, is ignored in this case and the models are “black box.” Since this is about movies, this makes a lot of sense!

Kaggle formed as a way to continuously host Netflix style competitions on the very toughest of data sets.