Reference

xgbtune.tune.tune_xgb_model(params, x_train, y_train, x_val=None, y_val=None, nfold=3, stratified=False, folds=None, shuffle=True, tune_params={}, max_round_count=5000, loss_compare=<built-in function lt>, pass_count=2, verbose=True)

Tunes a XGBoost model

Examples

>>> params, round_count = tune_xgb_model(x, y, x_val, y_val, model_params)

Parameters:

• params – A dictionary with the base xgboost parameters to use
• x_train – Train set
• y_train – Train labels
• x_val – Validation set
• y_val – Validation labels
• nfold – Number of folds for cv
• stratified – Perform stratified sampling
• folds – Sklearn KFolds or StratifiedKFolds object
• shuffle – shuffle data on cross validation
• tune_params – dictionary containing list of values to test to each parameter
• max_round_count – Maximum number of rounds during training
• pass_count – Number of tuning pass to do

Returns:

A tuple of tuned parameters and round count.