Machine Learning with Swift

XGBoost

When I started writing this book, I didn't expect to write about this tool. Why? Because this is a heavy artillery of machine learning contests. XGBoost is a production standard in many areas, but it is very resource consuming during the training phase: all your gigabytes will be consumed in the blink of an eye. That's why it is mainly used on servers and clusters of servers for web-ranking and other heavy-lifting tasks. It is also a tool that's considered to be a silver bullet for winning a Kaggle machine learning competition (if it's not about computer vision). CoreML supports gradient-boosting decision trees trained in XGBoost. For more information refer to: https://xgboost.readthedocs.io/en/latest/.