My capstone project for Udacity's Machine Learning Nanodegree
Topic: TalkingData AdTracking Fraud Detection Challenge
The training and testing datasets for the Fraud Detection Challenge can be downloaded from Kaggle's competition webpage.
Note: The notebook assumes data files are stored in ./data/
directory.
- Python >= 3.6
- numpy >= 1.14.3
- pandas >= 0.23.0
- scikit-learn >= 0.19.1
- xgboost == 0.72
Note: It is recommended to use XGBoost with GPU support for better performance. To enable GPU support for XGBoost, please build the library from source with flag -DUSE_CUDA=ON
. See more here.