Introduction I like large-scale machine learning and love to build scalable learning systems. XGBoost is one such project that we created. It is the package you want to use to solve your data-science problems. I might be a bit exaggerate, however, XGBoost indeed has been used by a series of kaggle winning solutions as well as KDDCup winners. It is fast and optimized for out-of-core computations. Y
![Story and Lessons Behind the Evolution of XGBoost](https://cdn-ak-scissors.b.st-hatena.com/image/square/d343a3cce4ff3f668ef7568f593e6f3719c138bc/height=288;version=1;width=512/http%3A%2F%2Fhomes.cs.washington.edu%2F~tqchen%2Fdata%2Fimg%2Fxgboost-pkg-stack.png)