The repository contains the final submission for a data science contest organized by PwC at Warsaw School of Economics. The objective was to create credit scoring model to estimate the default on loan probability of a given, potential borrower. To assess the contestants' models performane, Gini index was chosen. Thanks to the feature engineering, tuning the XGBoost model and proper test set error estimation, our model managed to score the highest AUC/Gini value among other contestants.

The main code, used to generate this submission, can be found in analysis.R file.

The final version of the presentation can be found in Group 106a - PwC Business Case-Competition 2019.pdf file.