CatBoost Algorithm Parameters

Data Science, AI and ML
#1

Training parameters from CatBoost:

  • loss_function the loss to be used for classification or regression;
  • eval_metric the model’s evaluation metric;
  • n_estimators the maximum number of decision trees;
  • learning_rate determines how fast or slow the model will learn;
  • depth the maximum depth for each tree;
  • ignored_features determines the features that should be ignored during training;
  • nan_mode the method that will be used to deal with missing values;
  • cat_features an array of categorical columns;
  • text_features for declaring text-based columns.