Xgboost 官方源码(python版本)

官网地址

一.XGBClassifier

xgboost.XGBClassifier(max_depth=3, 
                    learning_rate=0.1,
                    n_estimators=100, 
                    silent=True, 
                    objective='binary:logistic', 
                    booster='gbtree',
                    n_jobs=1, 
                    nthread=None,
                    gamma=0, 
                    min_child_weight=1, 
                    max_delta_step=0, 
                    subsample=1, 
                    colsample_bytree=1, 
                    colsample_bylevel=1, 
                    reg_alpha=0, 
                    reg_lambda=1, 
                    scale_pos_weight=1, 
                    base_score=0.5, 
                    random_state=0, 
                    seed=None, 
                    missing=None, 
                    **kwargs)

Parameters:

max_depth (int) – Maximum tree depth for base learners. learning_rate (float) – Boosting learning rate (xgb’s “eta”) n_estimators (int) – Number of boosted trees to fit. silent (boolean) – Whether to print messages while running boosting. objective (string) – Specify the learning task and the corresponding learning objective. Only “rank:pairwise” is supported currently. booster (string) – Specify which booster to use: gbtree, gblinear or dart. nthread (int) – Number of parallel threads used to run xgboost. (Deprecated, please use n_jobs) n_jobs (int) – Number of parallel threads used to run xgboost. (replaces nthread) gamma (float) – Minimum loss reduction required to make a further partition on a leaf node of the tree. min_child_weight (int) – Minimum sum of instance weight(hessian) needed in a child. max_delta_step (int) – Maximum delta step we allow each tree’s weight estimation to be. subsample (float) – Subsample ratio of the training instance. colsample_bytree (float) – Subsample ratio of columns when constructing each tree. colsample_bylevel (float) – Subsample ratio of columns for each split, in each level. reg_alpha (float (xgb's alpha)) – L1 regularization term on weights reg_lambda (float (xgb's lambda)) – L2 regularization term on weights scale_pos_weight (float) – Balancing of positive and negative weights. base_score – The initial prediction score of all instances, global bias. seed (int) – Random number seed. (Deprecated, please use random_state) random_state (int) – Random number seed. (replaces seed) missing (float, optional) – Value in the data which needs to be present as a missing value. If None, defaults to np.nan. **kwargs (dict, optional) – Keyword arguments for XGBoost Booster object. Full documentation of parameters can be found here: https://github.com/dmlc/xgboost/blob/master/doc/parameter.rst. Attempting to set a parameter via the constructor args and **kwargs dict simultaneously will result in a TypeError.

 

二.XGBRegressor

xgboost.XGBRegressor(max_depth=3,
                    learning_rate=0.1, 
                    n_estimators=100, 
                    silent=True, objective='reg:linear', 
                    booster='gbtree', 
                    n_jobs=1, 
                    nthread=None, 
                    gamma=0, 
                    min_child_weight=1, 
                    max_delta_step=0, 
                    subsample=1, 
                    colsample_bytree=1, 
                    colsample_bylevel=1, 
                    reg_alpha=0, 
                    reg_lambda=1, 
                    scale_pos_weight=1, 
                    base_score=0.5, 
                    random_state=0, 
                    seed=None, 
                    missing=None, 
                    importance_type='gain', 
                    **kwargs)

Parameters:

max_depth (int) – Maximum tree depth for base learners. learning_rate (float) – Boosting learning rate (xgb’s “eta”) n_estimators (int) – Number of boosted trees to fit. silent (boolean) – Whether to print messages while running boosting. objective (string or callable) – Specify the learning task and the corresponding learning objective or a custom objective function to be used (see note below). booster (string) – Specify which booster to use: gbtree, gblinear or dart. nthread (int) – Number of parallel threads used to run xgboost. (Deprecated, please use n_jobs) n_jobs (int) – Number of parallel threads used to run xgboost. (replaces nthread) gamma (float) – Minimum loss reduction required to make a further partition on a leaf node of the tree. min_child_weight (int) – Minimum sum of instance weight(hessian) needed in a child. max_delta_step (int) – Maximum delta step we allow each tree’s weight estimation to be. subsample (float) – Subsample ratio of the training instance. colsample_bytree (float) – Subsample ratio of columns when constructing each tree. colsample_bylevel (float) – Subsample ratio of columns for each split, in each level. reg_alpha (float (xgb's alpha)) – L1 regularization term on weights reg_lambda (float (xgb's lambda)) – L2 regularization term on weights scale_pos_weight (float) – Balancing of positive and negative weights. base_score – The initial prediction score of all instances, global bias. seed (int) – Random number seed. (Deprecated, please use random_state) random_state (int) – Random number seed. (replaces seed) missing (float, optional) – Value in the data which needs to be present as a missing value. If None, defaults to np.nan. importance_type (string, default "gain") – The feature importance type for the feature_importances_ property: either “gain”, “weight”, “cover”, “total_gain” or “total_cover”. **kwargs (dict, optional) – Keyword arguments for XGBoost Booster object. Full documentation of parameters can be found here: https://github.com/dmlc/xgboost/blob/master/doc/parameter.rst. Attempting to set a parameter via the constructor args and **kwargs dict simultaneously will result in a TypeError.