My final project's README

In this project we study how can improve an online prediction model's performance when training data and test data distributions are not the same.