These notebooks are my exploration of the Kaggle BNP Paribas Cardif Claims Kaggle Challenge. I've been using this challenge as a way to hone my statistics, machine learning, Python (w/ Pandas and Seaborn), and R skills. Here are some techniques I've been using:
- Gradient boosted random forests
- Pearson r coefficient to clean extraneous rows
- My own statistical test to check for reordering of categorical variables
- PCA
- Bayesian optimization of XGBoost parameters using BayesBoost
In this challenge I'm not super focused on getting a high ranking, but just on learning as much as I can.