xgboosttreenode Properties

XGBoost Tree© is an advanced implementation of a gradient boosting algorithm with a tree model as the base model. Boosting algorithms iteratively learn weak classifiers and then add them to a final strong classifier. XGBoost Tree is very flexible and provides many parameters that can be overwhelming to most users, so the XGBoost Tree node in SPSS® Modeler exposes the core features and commonly used parameters. The node is implemented in Python.

Table 1. xgboosttreenode properties
`xgboosttreenode` properties	Data type	Property description
`TargetField` Renamed to `target` starting with version 18.2.1.1	field	The target fields.
`InputFields` Renamed to `inputs` starting with version 18.2.1.1	field	The input fields.
`treeMethod` Renamed to `tree_method` starting with version 18.2.1.1	string	The tree method for model building. Possible values are `auto`, `exact`, or `approx`. Default is `auto`.
`numBoostRound` Renamed to `num_boost_round` starting with version 18.2.1.1	integer	The num boost round value for model building. Specify a value between `1` and `1000`. Default is `10`.
`maxDepth` Renamed to `max_depth` starting with version 18.2.1.1	integer	The max depth for tree growth. Specify a value of `1` or higher. Default is `6`.
`minChildWeight` Renamed to `min_child_weight` starting with version 18.2.1.1	Double	The min child weight for tree growth. Specify a value of `0` or higher. Default is `1`.
`maxDeltaStep` Renamed to `max_delta_step` starting with version 18.2.1.1	Double	The max delta step for tree growth. Specify a value of `0` or higher. Default is `0`.
`objectiveType` Renamed to `objective_type` starting with version 18.2.1.1	string	The objective type for the learning task. Possible values are `reg:linear`, `reg:logistic`, `reg:gamma`, `reg:tweedie`, `count:poisson`, `rank:pairwise`, `binary:logistic`, or `multi`. Note that for flag targets, only `binary:logistic` or `multi` can be used. If `multi` is used, the score result will show the `multi:softmax` and `multi:softprob` XGBoost objective types.
`earlyStopping` Renamed to `early_stopping` starting with version 18.2.1.1	Boolean	Whether to use the early stopping function. Default is `False`.
`earlyStoppingRounds` Renamed to `early_stopping_rounds` starting with version 18.2.1.1	integer	Validation error needs to decrease at least every early stopping round(s) to continue training. Default is `10`.
`evaluationDataRatio` Renamed to `evaluation_data_ratio` starting with version 18.2.1.1	Double	Ration of input data used for validation errors. Default is `0.3`.
`random_seed`	integer	The random number seed. Any number between `0` and `9999999`. Default is `0`.
`sampleSize` Renamed to `sample_size` starting with version 18.2.1.1	Double	The sub sample for control overfitting. Specify a value between `0.1` and `1.0`. Default is `0.1`.
`eta`	Double	The eta for control overfitting. Specify a value between `0` and `1`. Default is `0.3`.
`gamma`	Double	The gamma for control overfitting. Specify any number `0` or greater. Default is `6`.
`colsSampleRatio` Renamed to `col_sample_ratio` starting with version 18.2.1.1	Double	The colsample by tree for control overfitting. Specify a value between `0.01` and `1`. Default is `1`.
`colsSampleLevel` Renamed to `col_sample_level` starting with version 18.2.1.1	Double	The colsample by level for control overfitting. Specify a value between `0.01` and `1`. Default is `1`.
`lambda`	Double	The lambda for control overfitting. Specify any number `0` or greater. Default is `1`.
`alpha`	Double	The alpha for control overfitting. Specify any number `0` or greater. Default is `0`.
`scalePosWeight` Renamed to `scale_pos_weight` starting with version 18.2.1.1	Double	The scale pos weight for handling imbalanced datasets. Default is `1`.
`use_HPO` Added for version 18.2.1.1