Hands-On Machine Learning for Algorithmic Trading

How to tune hyperparameters

The numerous hyperparameters are listed in gbm_params.py. Each library has parameter settings to:

• Specify the overall objectives and learning algorithm
• Design the base learners
• Apply various regularization techniques
• Handle early stopping during training
• Enabling the use of GPU or parallelization on CPU

The documentation for each library details the various parameters that may refer to the same concept, but which have different names across libraries. The GitHub repository contains links to a site that highlights the corresponding parameters for xgboost and lightgbm.