GBTClassifier¶

class pyspark.ml.classification.GBTClassifier(*, featuresCol='features', labelCol='label', predictionCol='prediction', maxDepth=5, maxBins=32, minInstancesPerNode=1, minInfoGain=0.0, maxMemoryInMB=256, cacheNodeIds=False, checkpointInterval=10, lossType='logistic', maxIter=20, stepSize=0.1, seed=None, subsamplingRate=1.0, impurity='variance', featureSubsetStrategy='all', validationTol=0.01, validationIndicatorCol=None, leafCol='', minWeightFractionPerNode=0.0, weightCol=None)[source]¶

Gradient-Boosted Trees (GBTs) learning algorithm for classification. It supports binary labels, as well as both continuous and categorical features.

New in version 1.4.0.

Notes

Multiclass labels are not currently supported.

The implementation is based upon: J.H. Friedman. “Stochastic Gradient Boosting.” 1999.

Gradient Boosting vs. TreeBoost:

This implementation is for Stochastic Gradient Boosting, not for TreeBoost.
Both algorithms learn tree ensembles by minimizing loss functions.
TreeBoost (Friedman, 1999) additionally modifies the outputs at tree leaf nodes based on the loss function, whereas the original gradient boosting method does not.
We expect to implement TreeBoost in the future: SPARK-4240

Examples

>>> from numpy import allclose
>>> from pyspark.ml.linalg import Vectors
>>> from pyspark.ml.feature import StringIndexer
>>> df = spark.createDataFrame([
...     (1.0, Vectors.dense(1.0)),
...     (0.0, Vectors.sparse(1, [], []))], ["label", "features"])
>>> stringIndexer = StringIndexer(inputCol="label", outputCol="indexed")
>>> si_model = stringIndexer.fit(df)
>>> td = si_model.transform(df)
>>> gbt = GBTClassifier(maxIter=5, maxDepth=2, labelCol="indexed", seed=42,
...     leafCol="leafId")
>>> gbt.setMaxIter(5)
GBTClassifier...
>>> gbt.setMinWeightFractionPerNode(0.049)
GBTClassifier...
>>> gbt.getMaxIter()
5
>>> gbt.getFeatureSubsetStrategy()
'all'
>>> model = gbt.fit(td)
>>> model.getLabelCol()
'indexed'
>>> model.setFeaturesCol("features")
GBTClassificationModel...
>>> model.setThresholds([0.3, 0.7])
GBTClassificationModel...
>>> model.getThresholds()
[0.3, 0.7]
>>> model.featureImportances
SparseVector(1, {0: 1.0})
>>> allclose(model.treeWeights, [1.0, 0.1, 0.1, 0.1, 0.1])
True
>>> test0 = spark.createDataFrame([(Vectors.dense(-1.0),)], ["features"])
>>> model.predict(test0.head().features)
0.0
>>> model.predictRaw(test0.head().features)
DenseVector([1.1697, -1.1697])
>>> model.predictProbability(test0.head().features)
DenseVector([0.9121, 0.0879])
>>> result = model.transform(test0).head()
>>> result.prediction
0.0
>>> result.leafId
DenseVector([0.0, 0.0, 0.0, 0.0, 0.0])
>>> test1 = spark.createDataFrame([(Vectors.sparse(1, [0], [1.0]),)], ["features"])
>>> model.transform(test1).head().prediction
1.0
>>> model.totalNumNodes
15
>>> print(model.toDebugString)
GBTClassificationModel...numTrees=5...
>>> gbtc_path = temp_path + "gbtc"
>>> gbt.save(gbtc_path)
>>> gbt2 = GBTClassifier.load(gbtc_path)
>>> gbt2.getMaxDepth()
2
>>> model_path = temp_path + "gbtc_model"
>>> model.save(model_path)
>>> model2 = GBTClassificationModel.load(model_path)
>>> model.featureImportances == model2.featureImportances
True
>>> model.treeWeights == model2.treeWeights
True
>>> model.transform(test0).take(1) == model2.transform(test0).take(1)
True
>>> model.trees
[DecisionTreeRegressionModel...depth=..., DecisionTreeRegressionModel...]
>>> validation = spark.createDataFrame([(0.0, Vectors.dense(-1.0),)],
...              ["indexed", "features"])
>>> model.evaluateEachIteration(validation)
[0.25..., 0.23..., 0.21..., 0.19..., 0.18...]
>>> model.numClasses
2
>>> gbt = gbt.setValidationIndicatorCol("validationIndicator")
>>> gbt.getValidationIndicatorCol()
'validationIndicator'
>>> gbt.getValidationTol()
0.01

Methods

`clear`(param)	Clears a param from the param map if it has been explicitly set.
`copy`([extra])	Creates a copy of this instance with the same uid and some extra params.
`explainParam`(param)	Explains a single param and returns its name, doc, and optional default value and user-supplied value in a string.
`explainParams`()	Returns the documentation of all params with their optionally default values and user-supplied values.
`extractParamMap`([extra])	Extracts the embedded default param values and user-supplied values, and then merges them with extra values from input into a flat param map, where the latter value is used if there exist conflicts, i.e., with ordering: default param values < user-supplied values < extra.
`fit`(dataset[, params])	Fits a model to the input dataset with optional parameters.
`fitMultiple`(dataset, paramMaps)	Fits a model to the input dataset for each param map in paramMaps.
`getCacheNodeIds`()	Gets the value of cacheNodeIds or its default value.
`getCheckpointInterval`()	Gets the value of checkpointInterval or its default value.
`getFeatureSubsetStrategy`()	Gets the value of featureSubsetStrategy or its default value.
`getFeaturesCol`()	Gets the value of featuresCol or its default value.
`getImpurity`()	Gets the value of impurity or its default value.
`getLabelCol`()	Gets the value of labelCol or its default value.
`getLeafCol`()	Gets the value of leafCol or its default value.
`getLossType`()	Gets the value of lossType or its default value.
`getMaxBins`()	Gets the value of maxBins or its default value.
`getMaxDepth`()	Gets the value of maxDepth or its default value.
`getMaxIter`()	Gets the value of maxIter or its default value.
`getMaxMemoryInMB`()	Gets the value of maxMemoryInMB or its default value.
`getMinInfoGain`()	Gets the value of minInfoGain or its default value.
`getMinInstancesPerNode`()	Gets the value of minInstancesPerNode or its default value.
`getMinWeightFractionPerNode`()	Gets the value of minWeightFractionPerNode or its default value.
`getOrDefault`(param)	Gets the value of a param in the user-supplied param map or its default value.
`getParam`(paramName)	Gets a param by its name.
`getPredictionCol`()	Gets the value of predictionCol or its default value.
`getProbabilityCol`()	Gets the value of probabilityCol or its default value.
`getRawPredictionCol`()	Gets the value of rawPredictionCol or its default value.
`getSeed`()	Gets the value of seed or its default value.
`getStepSize`()	Gets the value of stepSize or its default value.
`getSubsamplingRate`()	Gets the value of subsamplingRate or its default value.
`getThresholds`()	Gets the value of thresholds or its default value.
`getValidationIndicatorCol`()	Gets the value of validationIndicatorCol or its default value.
`getValidationTol`()	Gets the value of validationTol or its default value.
`getWeightCol`()	Gets the value of weightCol or its default value.
`hasDefault`(param)	Checks whether a param has a default value.
`hasParam`(paramName)	Tests whether this instance contains a param with a given (string) name.
`isDefined`(param)	Checks whether a param is explicitly set by user or has a default value.
`isSet`(param)	Checks whether a param is explicitly set by user.
`load`(path)	Reads an ML instance from the input path, a shortcut of read().load(path).
`read`()	Returns an MLReader instance for this class.
`save`(path)	Save this ML instance to the given path, a shortcut of ‘write().save(path)’.
`set`(param, value)	Sets a parameter in the embedded param map.
`setCacheNodeIds`(value)	Sets the value of `cacheNodeIds`.
`setCheckpointInterval`(value)	Sets the value of `checkpointInterval`.
`setFeatureSubsetStrategy`(value)	Sets the value of `featureSubsetStrategy`.
`setFeaturesCol`(value)	Sets the value of `featuresCol`.
`setImpurity`(value)	Sets the value of `impurity`.
`setLabelCol`(value)	Sets the value of `labelCol`.
`setLeafCol`(value)	Sets the value of `leafCol`.
`setLossType`(value)	Sets the value of `lossType`.
`setMaxBins`(value)	Sets the value of `maxBins`.
`setMaxDepth`(value)	Sets the value of `maxDepth`.
`setMaxIter`(value)	Sets the value of `maxIter`.
`setMaxMemoryInMB`(value)	Sets the value of `maxMemoryInMB`.
`setMinInfoGain`(value)	Sets the value of `minInfoGain`.
`setMinInstancesPerNode`(value)	Sets the value of `minInstancesPerNode`.
`setMinWeightFractionPerNode`(value)	Sets the value of `minWeightFractionPerNode`.
`setParams`(self, \*[, featuresCol, labelCol, …])	Sets params for Gradient Boosted Tree Classification.
`setPredictionCol`(value)	Sets the value of `predictionCol`.
`setProbabilityCol`(value)	Sets the value of `probabilityCol`.
`setRawPredictionCol`(value)	Sets the value of `rawPredictionCol`.
`setSeed`(value)	Sets the value of `seed`.
`setStepSize`(value)	Sets the value of `stepSize`.
`setSubsamplingRate`(value)	Sets the value of `subsamplingRate`.
`setThresholds`(value)	Sets the value of `thresholds`.
`setValidationIndicatorCol`(value)	Sets the value of `validationIndicatorCol`.
`setWeightCol`(value)	Sets the value of `weightCol`.
`write`()	Returns an MLWriter instance for this ML instance.

Attributes

`cacheNodeIds`
`checkpointInterval`
`featureSubsetStrategy`
`featuresCol`
`impurity`
`labelCol`
`leafCol`
`lossType`
`maxBins`
`maxDepth`
`maxIter`
`maxMemoryInMB`
`minInfoGain`
`minInstancesPerNode`
`minWeightFractionPerNode`
`params`	Returns all params ordered by name.
`predictionCol`
`probabilityCol`
`rawPredictionCol`
`seed`
`stepSize`
`subsamplingRate`
`supportedFeatureSubsetStrategies`
`supportedImpurities`
`supportedLossTypes`
`thresholds`
`validationIndicatorCol`
`validationTol`
`weightCol`

Methods Documentation

clear(param)¶: Clears a param from the param map if it has been explicitly set.

copy(extra=None)¶

Creates a copy of this instance with the same uid and some extra params. This implementation first calls Params.copy and then make a copy of the companion Java pipeline component with extra params. So both the Python wrapper and the Java pipeline component get copied.

Parameters

extradict, optional: Extra parameters to copy to the new instance

Returns

JavaParams: Copy of this instance

explainParam(param)¶: Explains a single param and returns its name, doc, and optional default value and user-supplied value in a string.

explainParams()¶: Returns the documentation of all params with their optionally default values and user-supplied values.

extractParamMap(extra=None)¶

Extracts the embedded default param values and user-supplied values, and then merges them with extra values from input into a flat param map, where the latter value is used if there exist conflicts, i.e., with ordering: default param values < user-supplied values < extra.

Parameters

extradict, optional: extra param values

Returns

dict: merged param map

fit(dataset, params=None)¶

Fits a model to the input dataset with optional parameters.

New in version 1.3.0.

Parameters

datasetpyspark.sql.DataFrame: input dataset.
paramsdict or list or tuple, optional: an optional param map that overrides embedded params. If a list/tuple of param maps is given, this calls fit on each param map and returns a list of models.

Returns

Transformer or a list of Transformer: fitted model(s)

fitMultiple(dataset, paramMaps)¶

Fits a model to the input dataset for each param map in paramMaps.

New in version 2.3.0.

Parameters

datasetpyspark.sql.DataFrame: input dataset.
paramMapscollections.abc.Sequence: A Sequence of param maps.

Returns

_FitMultipleIterator: A thread safe iterable which contains one model for each param map. Each call to next(modelIterator) will return (index, model) where model was fit using paramMaps[index]. index values may not be sequential.

getCacheNodeIds()¶: Gets the value of cacheNodeIds or its default value.

getCheckpointInterval()¶: Gets the value of checkpointInterval or its default value.

getFeatureSubsetStrategy()¶: Gets the value of featureSubsetStrategy or its default value.

New in version 1.4.0.

getFeaturesCol()¶: Gets the value of featuresCol or its default value.

getImpurity()¶: Gets the value of impurity or its default value.

New in version 1.4.0.

getLabelCol()¶: Gets the value of labelCol or its default value.

getLeafCol()¶: Gets the value of leafCol or its default value.

getLossType()¶: Gets the value of lossType or its default value.

New in version 1.4.0.

getMaxBins()¶: Gets the value of maxBins or its default value.

getMaxDepth()¶: Gets the value of maxDepth or its default value.

getMaxIter()¶: Gets the value of maxIter or its default value.

getMaxMemoryInMB()¶: Gets the value of maxMemoryInMB or its default value.

getMinInfoGain()¶: Gets the value of minInfoGain or its default value.

getMinInstancesPerNode()¶: Gets the value of minInstancesPerNode or its default value.

getMinWeightFractionPerNode()¶: Gets the value of minWeightFractionPerNode or its default value.

getOrDefault(param)¶: Gets the value of a param in the user-supplied param map or its default value. Raises an error if neither is set.

getParam(paramName)¶: Gets a param by its name.

getPredictionCol()¶: Gets the value of predictionCol or its default value.

getProbabilityCol()¶: Gets the value of probabilityCol or its default value.

getRawPredictionCol()¶: Gets the value of rawPredictionCol or its default value.

getSeed()¶: Gets the value of seed or its default value.

getStepSize()¶: Gets the value of stepSize or its default value.

getSubsamplingRate()¶: Gets the value of subsamplingRate or its default value.

New in version 1.4.0.

getThresholds()¶: Gets the value of thresholds or its default value.

getValidationIndicatorCol()¶: Gets the value of validationIndicatorCol or its default value.

getValidationTol()¶: Gets the value of validationTol or its default value.

New in version 3.0.0.

getWeightCol()¶: Gets the value of weightCol or its default value.

hasDefault(param)¶: Checks whether a param has a default value.

hasParam(paramName)¶: Tests whether this instance contains a param with a given (string) name.

isDefined(param)¶: Checks whether a param is explicitly set by user or has a default value.

isSet(param)¶: Checks whether a param is explicitly set by user.

classmethod load(path)¶: Reads an ML instance from the input path, a shortcut of read().load(path).

classmethod read()¶: Returns an MLReader instance for this class.

save(path)¶: Save this ML instance to the given path, a shortcut of ‘write().save(path)’.

set(param, value)¶: Sets a parameter in the embedded param map.

setCacheNodeIds(value)[source]¶: Sets the value of cacheNodeIds.

setCheckpointInterval(value)[source]¶: Sets the value of checkpointInterval.

New in version 1.4.0.

setFeatureSubsetStrategy(value)[source]¶: Sets the value of featureSubsetStrategy.

New in version 2.4.0.

setFeaturesCol(value)¶: Sets the value of featuresCol.

New in version 3.0.0.

setImpurity(value)[source]¶: Sets the value of impurity.

New in version 1.4.0.

setLabelCol(value)¶: Sets the value of labelCol.

New in version 3.0.0.

setLeafCol(value)¶: Sets the value of leafCol.

setLossType(value)[source]¶: Sets the value of lossType.

New in version 1.4.0.

setMaxBins(value)[source]¶: Sets the value of maxBins.

setMaxDepth(value)[source]¶: Sets the value of maxDepth.

setMaxIter(value)[source]¶: Sets the value of maxIter.

New in version 1.4.0.

setMaxMemoryInMB(value)[source]¶: Sets the value of maxMemoryInMB.

setMinInfoGain(value)[source]¶: Sets the value of minInfoGain.

setMinInstancesPerNode(value)[source]¶: Sets the value of minInstancesPerNode.

setMinWeightFractionPerNode(value)[source]¶: Sets the value of minWeightFractionPerNode.

New in version 3.0.0.

setParams(self, \*, featuresCol="features", labelCol="label", predictionCol="prediction", maxDepth=5, maxBins=32, minInstancesPerNode=1, minInfoGain=0.0, maxMemoryInMB=256, cacheNodeIds=False, checkpointInterval=10, lossType="logistic", maxIter=20, stepSize=0.1, seed=None, subsamplingRate=1.0, impurity="variance", featureSubsetStrategy="all", validationTol=0.01, validationIndicatorCol=None, leafCol="", minWeightFractionPerNode=0.0, weightCol=None)[source]¶: Sets params for Gradient Boosted Tree Classification.

New in version 1.4.0.

setPredictionCol(value)¶: Sets the value of predictionCol.

New in version 3.0.0.

setProbabilityCol(value)¶: Sets the value of probabilityCol.

New in version 3.0.0.

setRawPredictionCol(value)¶: Sets the value of rawPredictionCol.

New in version 3.0.0.

setSeed(value)[source]¶: Sets the value of seed.

New in version 1.4.0.

setStepSize(value)[source]¶: Sets the value of stepSize.

New in version 1.4.0.

setSubsamplingRate(value)[source]¶: Sets the value of subsamplingRate.

New in version 1.4.0.

setThresholds(value)¶: Sets the value of thresholds.

New in version 3.0.0.

setValidationIndicatorCol(value)[source]¶: Sets the value of validationIndicatorCol.

New in version 3.0.0.

setWeightCol(value)[source]¶: Sets the value of weightCol.

New in version 3.0.0.

write()¶: Returns an MLWriter instance for this ML instance.

Attributes Documentation

cacheNodeIds = Param(parent='undefined', name='cacheNodeIds', doc='If false, the algorithm will pass trees to executors to match instances with nodes. If true, the algorithm will cache node IDs for each instance. Caching can speed up training of deeper trees. Users can set how often should the cache be checkpointed or disable it by setting checkpointInterval.')¶

checkpointInterval = Param(parent='undefined', name='checkpointInterval', doc='set checkpoint interval (>= 1) or disable checkpoint (-1). E.g. 10 means that the cache will get checkpointed every 10 iterations. Note: this setting will be ignored if the checkpoint directory is not set in the SparkContext.')¶

featureSubsetStrategy = Param(parent='undefined', name='featureSubsetStrategy', doc="The number of features to consider for splits at each tree node. Supported options: 'auto' (choose automatically for task: If numTrees == 1, set to 'all'. If numTrees > 1 (forest), set to 'sqrt' for classification and to 'onethird' for regression), 'all' (use all features), 'onethird' (use 1/3 of the features), 'sqrt' (use sqrt(number of features)), 'log2' (use log2(number of features)), 'n' (when n is in the range (0, 1.0], use n * number of features. When n is in the range (1, number of features), use n features). default = 'auto'")¶

featuresCol = Param(parent='undefined', name='featuresCol', doc='features column name.')¶

impurity = Param(parent='undefined', name='impurity', doc='Criterion used for information gain calculation (case-insensitive). Supported options: variance')¶

labelCol = Param(parent='undefined', name='labelCol', doc='label column name.')¶

leafCol = Param(parent='undefined', name='leafCol', doc='Leaf indices column name. Predicted leaf index of each instance in each tree by preorder.')¶

lossType = Param(parent='undefined', name='lossType', doc='Loss function which GBT tries to minimize (case-insensitive). Supported options: logistic')¶

maxBins = Param(parent='undefined', name='maxBins', doc='Max number of bins for discretizing continuous features. Must be >=2 and >= number of categories for any categorical feature.')¶

maxDepth = Param(parent='undefined', name='maxDepth', doc='Maximum depth of the tree. (>= 0) E.g., depth 0 means 1 leaf node; depth 1 means 1 internal node + 2 leaf nodes. Must be in range [0, 30].')¶

maxIter = Param(parent='undefined', name='maxIter', doc='max number of iterations (>= 0).')¶

maxMemoryInMB = Param(parent='undefined', name='maxMemoryInMB', doc='Maximum memory in MB allocated to histogram aggregation. If too small, then 1 node will be split per iteration, and its aggregates may exceed this size.')¶

minInfoGain = Param(parent='undefined', name='minInfoGain', doc='Minimum information gain for a split to be considered at a tree node.')¶

minInstancesPerNode = Param(parent='undefined', name='minInstancesPerNode', doc='Minimum number of instances each child must have after split. If a split causes the left or right child to have fewer than minInstancesPerNode, the split will be discarded as invalid. Should be >= 1.')¶

minWeightFractionPerNode = Param(parent='undefined', name='minWeightFractionPerNode', doc='Minimum fraction of the weighted sample count that each child must have after split. If a split causes the fraction of the total weight in the left or right child to be less than minWeightFractionPerNode, the split will be discarded as invalid. Should be in interval [0.0, 0.5).')¶

params¶: Returns all params ordered by name. The default implementation uses dir() to get all attributes of type Param.

predictionCol = Param(parent='undefined', name='predictionCol', doc='prediction column name.')¶

probabilityCol = Param(parent='undefined', name='probabilityCol', doc='Column name for predicted class conditional probabilities. Note: Not all models output well-calibrated probability estimates! These probabilities should be treated as confidences, not precise probabilities.')¶

rawPredictionCol = Param(parent='undefined', name='rawPredictionCol', doc='raw prediction (a.k.a. confidence) column name.')¶

seed = Param(parent='undefined', name='seed', doc='random seed.')¶

stepSize = Param(parent='undefined', name='stepSize', doc='Step size (a.k.a. learning rate) in interval (0, 1] for shrinking the contribution of each estimator.')¶

subsamplingRate = Param(parent='undefined', name='subsamplingRate', doc='Fraction of the training data used for learning each decision tree, in range (0, 1].')¶

supportedFeatureSubsetStrategies = ['auto', 'all', 'onethird', 'sqrt', 'log2']¶

supportedImpurities = ['variance']¶

supportedLossTypes = ['logistic']¶

thresholds = Param(parent='undefined', name='thresholds', doc="Thresholds in multi-class classification to adjust the probability of predicting each class. Array must have length equal to the number of classes, with values > 0, excepting that at most one value may be 0. The class with largest value p/t is predicted, where p is the original probability of that class and t is the class's threshold.")¶

validationIndicatorCol = Param(parent='undefined', name='validationIndicatorCol', doc='name of the column that indicates whether each row is for training or for validation. False indicates training; true indicates validation.')¶

validationTol = Param(parent='undefined', name='validationTol', doc='Threshold for stopping early when fit with validation is used. If the error rate on the validation input changes by less than the validationTol, then learning will stop early (before `maxIter`). This parameter is ignored when fit without validation is used.')¶

weightCol = Param(parent='undefined', name='weightCol', doc='weight column name. If this is not set or empty, we treat all instance weights as 1.0.')¶

DecisionTreeClassificationModel GBTClassificationModel