Recently I was working on a in-class competition from the “How to win a data science competition” Coursera course. You can start for free with the 7-day Free Trial. Learned a lot of new things from that about using XGBoost for time series prediction tasks. The one thing that I tried out in this competition was the Hyperopt package - A bayesian Parameter Tuning Framework. And I was literally amazed.
THE PROBLEM: Recently I was working on the Criteo Advertising Competition on Kaggle. The competition was a classification problem which basically involved predicting the click through rates based on several features provided in the train data. Seeing the size of the data (11 GB Train), I felt that going with Vowpal Wabbit might be a better option. But after getting to an CV error of .47 on the Kaggle LB and being stuck there , I felt the need to go back to Scikit learn.