XGBoost is an open source, ensemble machine learning algorithm that utilizes a high-performance implementation of gradient boosted decision trees. An underlying C++ codebase combined with a Python interface sitting on top makes XGBoost a very fast, scalable, and highly usable library.