def __init__(self, keep=False, columns=[]):
super(ColumnPruner, self).__init__()
self._java_obj = self._new_java_obj(
"py_sparkling.ml.features.ColumnPruner", self.uid)
self._setDefault(keep=False, columns=[])
kwargs = get_input_kwargs(self)
self.setParams(**kwargs)
def __init__(self,
foldCol=None,
labelCol="label",
inputCols=[],
holdoutStrategy="None",
blendedAvgEnabled=False,
blendedAvgInflectionPoint=10.0,
blendedAvgSmoothing=20.0,
noise=0.01,
noiseSeed=-1):
super(H2OTargetEncoder, self).__init__()
self._hc = H2OContext.getOrCreate(SparkSession.builder.getOrCreate(),
verbose=False)
self._java_obj = self._new_java_obj(
"ai.h2o.sparkling.ml.features.H2OTargetEncoder", self.uid)
self._setDefault(foldCol=None,
labelCol="label",
inputCols=[],
holdoutStrategy="None",
blendedAvgEnabled=False,
blendedAvgInflectionPoint=10.0,
blendedAvgSmoothing=20.0,
noise=0.01,
noiseSeed=-1)
kwargs = get_input_kwargs(self)
self.setParams(**kwargs)
def __init__(self,
labelCol="label",
allStringColumnsToCategorical=True,
columnsToCategorical=[],
ratio=1.0,
foldCol=None,
weightCol=None,
ignoredCols=[],
includeAlgos=None,
excludeAlgos=None,
projectName=None,
maxRuntimeSecs=3600.0,
stoppingRounds=3,
stoppingTolerance=0.001,
stoppingMetric="AUTO",
nfolds=5,
convertUnknownCategoricalLevelsToNa=False,
seed=-1,
sortMetric="AUTO",
balanceClasses=False,
classSamplingFactors=None,
maxAfterBalanceSize=5.0,
keepCrossValidationPredictions=True,
keepCrossValidationModels=True,
maxModels=0,
**deprecatedArgs):
super(H2OAutoML, self).__init__()
self._hc = H2OContext.getOrCreate(SparkSession.builder.getOrCreate(),
verbose=False)
self._java_obj = self._new_java_obj("py_sparkling.ml.algos.H2OAutoML",
self.uid)
self._setDefault(labelCol="label",
allStringColumnsToCategorical=True,
columnsToCategorical=[],
ratio=1.0,
foldCol=None,
weightCol=None,
ignoredCols=[],
includeAlgos=None,
excludeAlgos=None,
projectName=None,
maxRuntimeSecs=3600.0,
stoppingRounds=3,
stoppingTolerance=0.001,
stoppingMetric=self._hc._jvm.hex.ScoreKeeper.
StoppingMetric.valueOf("AUTO"),
nfolds=5,
convertUnknownCategoricalLevelsToNa=False,
seed=-1,
sortMetric=None,
balanceClasses=False,
classSamplingFactors=None,
maxAfterBalanceSize=5.0,
keepCrossValidationPredictions=True,
keepCrossValidationModels=True,
maxModels=0)
kwargs = get_input_kwargs(self)
self.setParams(**kwargs)
def setParams(self,
modelId=None,
splitRatio=1.0,
labelCol="label",
weightCol=None,
featuresCols=[],
allStringColumnsToCategorical=True,
columnsToCategorical=[],
nfolds=0,
keepCrossValidationPredictions=False,
keepCrossValidationFoldAssignment=False,
parallelizeCrossValidation=True,
seed=-1,
distribution="AUTO",
epochs=10.0,
l1=0.0,
l2=0.0,
hidden=[200, 200],
reproducible=False,
convertUnknownCategoricalLevelsToNa=False,
foldCol=None,
**deprecatedArgs):
kwargs = get_input_kwargs(self)
if "distribution" in kwargs:
kwargs[
"distribution"] = self._hc._jvm.hex.genmodel.utils.DistributionFamily.valueOf(
kwargs["distribution"])
# we need to convert double arguments manually to floats as if we assign integer to double, py4j thinks that
# the whole type is actually int and we get class cast exception
double_types = ["splitRatio", "epochs", "l1", "l2"]
set_double_values(kwargs, double_types)
return self._set(**kwargs)
def setParams(self,
featuresCols=[],
labelCol="label",
allStringColumnsToCategorical=True,
columnsToCategorical=[],
splitRatio=1.0,
foldCol=None,
weightCol=None,
ignoredCols=[],
includeAlgos=None,
excludeAlgos=None,
projectName=None,
maxRuntimeSecs=3600.0,
stoppingRounds=3,
stoppingTolerance=0.001,
stoppingMetric="AUTO",
nfolds=5,
convertUnknownCategoricalLevelsToNa=True,
seed=-1,
sortMetric="AUTO",
balanceClasses=False,
classSamplingFactors=None,
maxAfterBalanceSize=5.0,
keepCrossValidationPredictions=True,
keepCrossValidationModels=True,
maxModels=0,
**deprecatedArgs):
kwargs = get_input_kwargs(self)
if "stoppingMetric" in kwargs:
kwargs[
"stoppingMetric"] = self._hc._jvm.hex.ScoreKeeper.StoppingMetric.valueOf(
kwargs["stoppingMetric"])
if "projectName" in kwargs and kwargs["projectName"] is None:
kwargs["projectName"] = ''.join(
random.choice(string.ascii_letters) for i in range(30))
if "excludeAlgos" in kwargs:
jvm = H2OContext.getOrCreate(SparkSession.builder.getOrCreate(),
verbose=False)._jvm
kwargs["excludeAlgos"] = get_enum_array_from_str_array(
kwargs["excludeAlgos"], jvm.ai.h2o.automl.Algo)
if "includeAlgos" in kwargs:
jvm = H2OContext.getOrCreate(SparkSession.builder.getOrCreate(),
verbose=False)._jvm
kwargs["includeAlgos"] = get_enum_array_from_str_array(
kwargs["includeAlgos"], jvm.ai.h2o.automl.Algo)
# we need to convert double arguments manually to floats as if we assign integer to double, py4j thinks that
# the whole type is actually int and we get class cast exception
double_types = [
"maxRuntimeSecs", "stoppingTolerance", "splitRatio",
"maxAfterBalanceSize"
]
set_double_values(kwargs, double_types)
return self._set(**kwargs)
def setParams(self,
featuresCols=[],
labelCol="label",
allStringColumnsToCategorical=True,
columnsToCategorical=[],
splitRatio=1.0,
foldCol=None,
weightCol=None,
ignoredCols=[],
includeAlgos=None,
excludeAlgos=None,
projectName=None,
maxRuntimeSecs=3600.0,
stoppingRounds=3,
stoppingTolerance=0.001,
stoppingMetric="AUTO",
nfolds=5,
convertUnknownCategoricalLevelsToNa=True,
seed=-1,
sortMetric="AUTO",
balanceClasses=False,
classSamplingFactors=None,
maxAfterBalanceSize=5.0,
keepCrossValidationPredictions=True,
keepCrossValidationModels=True,
maxModels=0,
predictionCol="prediction",
detailedPredictionCol="detailed_prediction",
withDetailedPredictionCol=False,
convertInvalidNumbersToNa=False,
**deprecatedArgs):
kwargs = get_input_kwargs(self)
validateEnumValues(self._H2OAutoMLParams__getAutomlAlgoEnum(),
kwargs,
"includeAlgos",
nullEnabled=True)
validateEnumValues(self._H2OAutoMLParams__getAutomlAlgoEnum(),
kwargs,
"excludeAlgos",
nullEnabled=True)
validateEnumValue(self._H2OAutoMLParams__getStoppingMetricEnum(),
kwargs, "stoppingMetric")
if "projectName" in kwargs and kwargs["projectName"] is None:
kwargs["projectName"] = ''.join(
random.choice(string.ascii_letters) for i in range(30))
# we need to convert double arguments manually to floats as if we assign integer to double, py4j thinks that
# the whole type is actually int and we get class cast exception
double_types = [
"maxRuntimeSecs", "stoppingTolerance", "splitRatio",
"maxAfterBalanceSize"
]
set_double_values(kwargs, double_types)
return self._set(**kwargs)
def __init__(self,
modelId=None,
splitRatio=1.0,
labelCol="label",
weightCol=None,
featuresCols=[],
allStringColumnsToCategorical=True,
columnsToCategorical=[],
nfolds=0,
keepCrossValidationPredictions=False,
keepCrossValidationFoldAssignment=False,
parallelizeCrossValidation=True,
seed=-1,
distribution="AUTO",
epochs=10.0,
l1=0.0,
l2=0.0,
hidden=[200, 200],
reproducible=False,
convertUnknownCategoricalLevelsToNa=False,
foldCol=None,
predictionCol="prediction",
detailedPredictionCol="detailed_prediction",
withDetailedPredictionCol=False,
convertInvalidNumbersToNa=False,
**deprecatedArgs):
super(H2ODeepLearning, self).__init__()
self._java_obj = self._new_java_obj(
"py_sparkling.ml.algos.H2ODeepLearning", self.uid)
self._setDefault(modelId=None,
splitRatio=1.0,
labelCol="label",
weightCol=None,
featuresCols=[],
allStringColumnsToCategorical=True,
columnsToCategorical=[],
nfolds=0,
keepCrossValidationPredictions=False,
keepCrossValidationFoldAssignment=False,
parallelizeCrossValidation=True,
seed=-1,
distribution="AUTO",
epochs=10.0,
l1=0.0,
l2=0.0,
hidden=[200, 200],
reproducible=False,
convertUnknownCategoricalLevelsToNa=False,
foldCol=None,
predictionCol="prediction",
detailedPredictionCol="detailed_prediction",
withDetailedPredictionCol=False,
convertInvalidNumbersToNa=False)
kwargs = get_input_kwargs(self)
self.setParams(**kwargs)
def __init__(self,
featuresCols=[],
algo=None,
splitRatio=1.0,
hyperParameters={},
labelCol="label",
weightCol=None,
allStringColumnsToCategorical=True,
columnsToCategorical=[],
strategy="Cartesian",
maxRuntimeSecs=0.0,
maxModels=0,
seed=-1,
stoppingRounds=0,
stoppingTolerance=0.001,
stoppingMetric="AUTO",
nfolds=0,
selectBestModelBy="AUTO",
selectBestModelDecreasing=True,
foldCol=None,
convertUnknownCategoricalLevelsToNa=True,
predictionCol="prediction",
detailedPredictionCol="detailed_prediction",
withDetailedPredictionCol=False,
convertInvalidNumbersToNa=False,
**deprecatedArgs):
super(H2OGridSearch, self).__init__()
self._java_obj = self._new_java_obj(
"py_sparkling.ml.algos.H2OGridSearch", self.uid)
self._setDefault(featuresCols=[],
algo=None,
splitRatio=1.0,
hyperParameters={},
labelCol="label",
weightCol=None,
allStringColumnsToCategorical=True,
columnsToCategorical=[],
strategy="Cartesian",
maxRuntimeSecs=0.0,
maxModels=0,
seed=-1,
stoppingRounds=0,
stoppingTolerance=0.001,
stoppingMetric="AUTO",
nfolds=0,
selectBestModelBy="AUTO",
selectBestModelDecreasing=True,
foldCol=None,
convertUnknownCategoricalLevelsToNa=True,
predictionCol="prediction",
detailedPredictionCol="detailed_prediction",
withDetailedPredictionCol=False,
convertInvalidNumbersToNa=False)
kwargs = get_input_kwargs(self)
self.setParams(**kwargs)
def __init__(self,
predictionCol="prediction",
detailedPredictionCol="detailed_prediction",
withDetailedPredictionCol=False,
featuresCols=[],
foldCol=None,
weightCol=None,
splitRatio=1.0,
seed=-1,
nfolds=0,
allStringColumnsToCategorical=True,
columnsToCategorical=[],
convertUnknownCategoricalLevelsToNa=False,
convertInvalidNumbersToNa=False,
modelId=None,
keepCrossValidationPredictions=False,
keepCrossValidationFoldAssignment=False,
parallelizeCrossValidation=True,
distribution="AUTO",
maxIterations=10,
standardize=True,
init="Furthest",
userPoints=None,
estimateK=False,
k=2,
**deprecatedArgs):
super(H2OKMeans, self).__init__()
self._java_obj = self._new_java_obj(
"ai.h2o.sparkling.ml.algos.H2OKMeans", self.uid)
self._setDefault(predictionCol="prediction",
detailedPredictionCol="detailed_prediction",
withDetailedPredictionCol=False,
featuresCols=[],
foldCol=None,
weightCol=None,
splitRatio=1.0,
seed=-1,
nfolds=0,
allStringColumnsToCategorical=True,
columnsToCategorical=[],
convertUnknownCategoricalLevelsToNa=False,
convertInvalidNumbersToNa=False,
modelId=None,
keepCrossValidationPredictions=False,
keepCrossValidationFoldAssignment=False,
parallelizeCrossValidation=True,
distribution="AUTO",
maxIterations=10,
standardize=True,
init="Furthest",
userPoints=None,
estimateK=False,
k=2)
kwargs = get_input_kwargs(self)
self.setParams(**kwargs)
def setParams(self,
featuresCols=[],
algo=None,
splitRatio=1.0,
hyperParameters={},
labelCol="label",
weightCol=None,
allStringColumnsToCategorical=True,
columnsToCategorical=[],
strategy="Cartesian",
maxRuntimeSecs=0.0,
maxModels=0,
seed=-1,
stoppingRounds=0,
stoppingTolerance=0.001,
stoppingMetric="AUTO",
nfolds=0,
selectBestModelBy=None,
selectBestModelDecreasing=True,
foldCol=None,
convertUnknownCategoricalLevelsToNa=True,
**deprecatedArgs):
kwargs = get_input_kwargs(self)
if "stoppingMetric" in kwargs:
kwargs[
"stoppingMetric"] = self._hc._jvm.hex.ScoreKeeper.StoppingMetric.valueOf(
kwargs["stoppingMetric"])
if "strategy" in kwargs:
kwargs[
"strategy"] = self._hc._jvm.hex.grid.HyperSpaceSearchCriteria.Strategy.valueOf(
kwargs["strategy"])
if "selectBestModelBy" in kwargs and kwargs[
"selectBestModelBy"] is not None:
kwargs[
"selectBestModelBy"] = self._hc._jvm.org.apache.spark.ml.h2o.algos.H2OGridSearchMetric.valueOf(
kwargs["selectBestModelBy"])
propagate_value_from_deprecated_property(kwargs, "predictionCol",
"labelCol")
propagate_value_from_deprecated_property(kwargs, "ratio", "splitRatio")
# we need to convert double arguments manually to floats as if we assign integer to double, py4j thinks that
# the whole type is actually int and we get class cast exception
double_types = ["ratio", "stoppingTolerance", "maxRuntimeSecs"]
set_double_values(kwargs, double_types)
if "algo" in kwargs and kwargs["algo"] is not None:
tmp = kwargs["algo"]
del kwargs['algo']
self._java_obj.setAlgo(tmp._java_obj)
return self._set(**kwargs)
def __init__(self,
featuresCols=[],
algo=None,
splitRatio=1.0,
hyperParameters={},
labelCol="label",
weightCol=None,
allStringColumnsToCategorical=True,
columnsToCategorical=[],
strategy="Cartesian",
maxRuntimeSecs=0.0,
maxModels=0,
seed=-1,
stoppingRounds=0,
stoppingTolerance=0.001,
stoppingMetric="AUTO",
nfolds=0,
selectBestModelBy=None,
selectBestModelDecreasing=True,
foldCol=None,
convertUnknownCategoricalLevelsToNa=True,
**deprecatedArgs):
super(H2OGridSearch, self).__init__()
self._hc = H2OContext.getOrCreate(SparkSession.builder.getOrCreate(),
verbose=False)
self._java_obj = self._new_java_obj(
"py_sparkling.ml.algos.H2OGridSearch", self.uid)
self._setDefault(
featuresCols=[],
algo=None,
splitRatio=1.0,
hyperParameters={},
labelCol="label",
weightCol=None,
allStringColumnsToCategorical=True,
columnsToCategorical=[],
strategy=self._hc._jvm.hex.grid.HyperSpaceSearchCriteria.Strategy.
valueOf("Cartesian"),
maxRuntimeSecs=0.0,
maxModels=0,
seed=-1,
stoppingRounds=0,
stoppingTolerance=0.001,
stoppingMetric=self._hc._jvm.hex.ScoreKeeper.StoppingMetric.
valueOf("AUTO"),
nfolds=0,
selectBestModelBy=None,
selectBestModelDecreasing=True,
foldCol=None,
convertUnknownCategoricalLevelsToNa=True)
kwargs = get_input_kwargs(self)
self.setParams(**kwargs)
def __init__(self,
modelId=None,
splitRatio=1.0,
labelCol="label",
weightCol=None,
featuresCols=[],
allStringColumnsToCategorical=True,
columnsToCategorical=[],
nfolds=0,
keepCrossValidationPredictions=False,
keepCrossValidationFoldAssignment=False,
parallelizeCrossValidation=True,
seed=-1,
distribution="AUTO",
epochs=10.0,
l1=0.0,
l2=0.0,
hidden=[200, 200],
reproducible=False,
convertUnknownCategoricalLevelsToNa=False,
foldCol=None,
**deprecatedArgs):
super(H2ODeepLearning, self).__init__()
self._hc = H2OContext.getOrCreate(SparkSession.builder.getOrCreate(),
verbose=False)
self._java_obj = self._new_java_obj(
"py_sparkling.ml.algos.H2ODeepLearning", self.uid)
self._setDefault(modelId=None,
splitRatio=1.0,
labelCol="label",
weightCol=None,
featuresCols=[],
allStringColumnsToCategorical=True,
columnsToCategorical=[],
nfolds=0,
keepCrossValidationPredictions=False,
keepCrossValidationFoldAssignment=False,
parallelizeCrossValidation=True,
seed=-1,
distribution=self._hc._jvm.hex.genmodel.utils.
DistributionFamily.valueOf("AUTO"),
epochs=10.0,
l1=0.0,
l2=0.0,
hidden=[200, 200],
reproducible=False,
convertUnknownCategoricalLevelsToNa=False,
foldCol=None)
kwargs = get_input_kwargs(self)
self.setParams(**kwargs)
def setParams(self,
featuresCols=[],
algo=None,
splitRatio=1.0,
hyperParameters={},
labelCol="label",
weightCol=None,
allStringColumnsToCategorical=True,
columnsToCategorical=[],
strategy="Cartesian",
maxRuntimeSecs=0.0,
maxModels=0,
seed=-1,
stoppingRounds=0,
stoppingTolerance=0.001,
stoppingMetric="AUTO",
nfolds=0,
selectBestModelBy="AUTO",
selectBestModelDecreasing=True,
foldCol=None,
convertUnknownCategoricalLevelsToNa=True,
predictionCol="prediction",
detailedPredictionCol="detailed_prediction",
withDetailedPredictionCol=False,
convertInvalidNumbersToNa=False,
**deprecatedArgs):
kwargs = get_input_kwargs(self)
validateEnumValue(self._H2OGridSearchParams__getStrategyEnum(), kwargs,
"strategy")
validateEnumValue(self._H2OGridSearchParams__getStoppingMetricEnum(),
kwargs, "stoppingMetric")
validateEnumValue(
self._H2OGridSearchParams__getSelectBestModelByEnum(), kwargs,
"selectBestModelBy")
# we need to convert double arguments manually to floats as if we assign integer to double, py4j thinks that
# the whole type is actually int and we get class cast exception
double_types = ["splitRatio", "stoppingTolerance", "maxRuntimeSecs"]
set_double_values(kwargs, double_types)
if "algo" in kwargs and kwargs["algo"] is not None:
tmp = kwargs["algo"]
del kwargs['algo']
self._java_obj.setAlgo(tmp._java_obj)
return self._set(**kwargs)
def setParams(self,
predictionCol="prediction",
detailedPredictionCol="detailed_prediction",
withDetailedPredictionCol=False,
featuresCols=[],
foldCol=None,
weightCol=None,
splitRatio=1.0,
seed=-1,
nfolds=0,
allStringColumnsToCategorical=True,
columnsToCategorical=[],
convertUnknownCategoricalLevelsToNa=False,
convertInvalidNumbersToNa=False,
modelId=None,
keepCrossValidationPredictions=False,
keepCrossValidationFoldAssignment=False,
parallelizeCrossValidation=True,
distribution="AUTO",
maxIterations=10,
standardize=True,
init="Furthest",
userPoints=None,
estimateK=False,
k=2,
**deprecatedArgs):
kwargs = get_input_kwargs(self)
validateEnumValue(self._H2OAlgoCommonParams__getDistributionEnum(),
kwargs, "distribution")
validateEnumValue(self._H2OKMeansParams__getInitEnum(), kwargs, "init")
# We need to convert double arguments manually to floats as if we assign integer to double, py4j thinks that
# the whole type is actually int and we get class cast exception
double_types = ["splitRatio"]
set_double_values(kwargs, double_types)
if "init" in kwargs:
kwargs["init"] = getDoubleArrayArrayFromIntArrayArray(
kwargs["init"])
return self._set(**kwargs)
def setParams(self,
foldCol=None,
labelCol="label",
inputCols=[],
holdoutStrategy="None",
blendedAvgEnabled=False,
blendedAvgInflectionPoint=10.0,
blendedAvgSmoothing=20.0,
noise=0.01,
noiseSeed=-1):
kwargs = get_input_kwargs(self)
validateEnumValue(self.__getHoldoutStrategyEnumName(), kwargs,
"holdoutStrategy")
# we need to convert double arguments manually to floats as if we assign integer to double, py4j thinks that
# the whole type is actually int and we get class cast exception
double_types = [
"blendedAvgInflectionPoint", "blendedAvgSmoothing", "noise"
]
set_double_values(kwargs, double_types)
return self._set(**kwargs)
def setParams(self,
modelId=None,
splitRatio=1.0,
labelCol="label",
weightCol=None,
featuresCols=[],
allStringColumnsToCategorical=True,
columnsToCategorical=[],
nfolds=0,
keepCrossValidationPredictions=False,
keepCrossValidationFoldAssignment=False,
parallelizeCrossValidation=True,
seed=-1,
distribution="AUTO",
epochs=10.0,
l1=0.0,
l2=0.0,
hidden=[200, 200],
reproducible=False,
convertUnknownCategoricalLevelsToNa=False,
foldCol=None,
predictionCol="prediction",
detailedPredictionCol="detailed_prediction",
withDetailedPredictionCol=False,
convertInvalidNumbersToNa=False,
**deprecatedArgs):
kwargs = get_input_kwargs(self)
validateEnumValue(self._H2OAlgoCommonParams__getDistributionEnum(),
kwargs, "distribution")
# we need to convert double arguments manually to floats as if we assign integer to double, py4j thinks that
# the whole type is actually int and we get class cast exception
double_types = ["splitRatio", "epochs", "l1", "l2"]
set_double_values(kwargs, double_types)
return self._set(**kwargs)
def __init__(self,
modelId=None,
splitRatio=1.0,
labelCol="label",
weightCol=None,
featuresCols=[],
allStringColumnsToCategorical=True,
columnsToCategorical=[],
nfolds=0,
keepCrossValidationPredictions=False,
keepCrossValidationFoldAssignment=False,
parallelizeCrossValidation=True,
seed=-1,
distribution="AUTO",
convertUnknownCategoricalLevelsToNa=False,
quietMode=True,
missingValuesHandling=None,
ntrees=50,
nEstimators=0,
maxDepth=6,
minRows=1.0,
minChildWeight=1.0,
learnRate=0.3,
eta=0.3,
learnRateAnnealing=1.0,
sampleRate=1.0,
subsample=1.0,
colSampleRate=1.0,
colSampleByLevel=1.0,
colSampleRatePerTree=1.0,
colsampleBytree=1.0,
maxAbsLeafnodePred=0.0,
maxDeltaStep=0.0,
scoreTreeInterval=0,
initialScoreInterval=4000,
scoreInterval=4000,
minSplitImprovement=0.0,
gamma=0.0,
nthread=-1,
maxBins=256,
maxLeaves=0,
minSumHessianInLeaf=100.0,
minDataInLeaf=0.0,
treeMethod="auto",
growPolicy="depthwise",
booster="gbtree",
dmatrixType="auto",
regLambda=0.0,
regAlpha=0.0,
sampleType="uniform",
normalizeType="tree",
rateDrop=0.0,
oneDrop=False,
skipDrop=0.0,
gpuId=0,
backend="auto",
foldCol=None,
**deprecatedArgs):
super(H2OXGBoost, self).__init__()
self._hc = H2OContext.getOrCreate(SparkSession.builder.getOrCreate(),
verbose=False)
self._java_obj = self._new_java_obj("py_sparkling.ml.algos.H2OXGBoost",
self.uid)
self._setDefault(
modelId=None,
splitRatio=1.0,
labelCol="label",
weightCol=None,
featuresCols=[],
allStringColumnsToCategorical=True,
columnsToCategorical=[],
nfolds=0,
keepCrossValidationPredictions=False,
keepCrossValidationFoldAssignment=False,
parallelizeCrossValidation=True,
seed=-1,
distribution=self._hc._jvm.hex.genmodel.utils.DistributionFamily.
valueOf("AUTO"),
convertUnknownCategoricalLevelsToNa=False,
quietMode=True,
missingValuesHandling=None,
ntrees=50,
nEstimators=0,
maxDepth=6,
minRows=1.0,
minChildWeight=1.0,
learnRate=0.3,
eta=0.3,
learnRateAnnealing=1.0,
sampleRate=1.0,
subsample=1.0,
colSampleRate=1.0,
colSampleByLevel=1.0,
colSampleRatePerTree=1.0,
colsampleBytree=1.0,
maxAbsLeafnodePred=0.0,
maxDeltaStep=0.0,
scoreTreeInterval=0,
initialScoreInterval=4000,
scoreInterval=4000,
minSplitImprovement=0.0,
gamma=0.0,
nthread=-1,
maxBins=256,
maxLeaves=0,
minSumHessianInLeaf=100.0,
minDataInLeaf=0.0,
treeMethod=self._hc._jvm.hex.tree.xgboost.XGBoostModel.
XGBoostParameters.TreeMethod.valueOf("auto"),
growPolicy=self._hc._jvm.hex.tree.xgboost.XGBoostModel.
XGBoostParameters.GrowPolicy.valueOf("depthwise"),
booster=self._hc._jvm.hex.tree.xgboost.XGBoostModel.
XGBoostParameters.Booster.valueOf("gbtree"),
dmatrixType=self._hc._jvm.hex.tree.xgboost.XGBoostModel.
XGBoostParameters.DMatrixType.valueOf("auto"),
regLambda=0.0,
regAlpha=0.0,
sampleType=self._hc._jvm.hex.tree.xgboost.XGBoostModel.
XGBoostParameters.DartSampleType.valueOf("uniform"),
normalizeType=self._hc._jvm.hex.tree.xgboost.XGBoostModel.
XGBoostParameters.DartNormalizeType.valueOf("tree"),
rateDrop=0.0,
oneDrop=False,
skipDrop=0.0,
gpuId=0,
backend=self._hc._jvm.hex.tree.xgboost.XGBoostModel.
XGBoostParameters.Backend.valueOf("auto"),
foldCol=None)
kwargs = get_input_kwargs(self)
self.setParams(**kwargs)
def __init__(self,
modelId=None,
splitRatio=1.0,
labelCol="label",
weightCol=None,
featuresCols=[],
allStringColumnsToCategorical=True,
columnsToCategorical=[],
nfolds=0,
keepCrossValidationPredictions=False,
keepCrossValidationFoldAssignment=False,
parallelizeCrossValidation=True,
seed=-1,
distribution="AUTO",
ntrees=50,
maxDepth=5,
minRows=10.0,
nbins=20,
nbinsCats=1024,
minSplitImprovement=1e-5,
histogramType="AUTO",
r2Stopping=java_max_double_value,
nbinsTopLevel=1 << 10,
buildTreeOneNode=False,
scoreTreeInterval=0,
sampleRate=1.0,
sampleRatePerClass=None,
colSampleRateChangePerLevel=1.0,
colSampleRatePerTree=1.0,
learnRate=0.1,
learnRateAnnealing=1.0,
colSampleRate=1.0,
maxAbsLeafnodePred=java_max_double_value,
predNoiseBandwidth=0.0,
convertUnknownCategoricalLevelsToNa=False,
foldCol=None,
**deprecatedArgs):
super(H2OGBM, self).__init__()
self._hc = H2OContext.getOrCreate(SparkSession.builder.getOrCreate(),
verbose=False)
self._java_obj = self._new_java_obj("py_sparkling.ml.algos.H2OGBM",
self.uid)
self._setDefault(modelId=None,
splitRatio=1.0,
labelCol="label",
weightCol=None,
featuresCols=[],
allStringColumnsToCategorical=True,
columnsToCategorical=[],
nfolds=0,
keepCrossValidationPredictions=False,
keepCrossValidationFoldAssignment=False,
parallelizeCrossValidation=True,
seed=-1,
distribution=self._hc._jvm.hex.genmodel.utils.
DistributionFamily.valueOf("AUTO"),
ntrees=50,
maxDepth=5,
minRows=10.0,
nbins=20,
nbinsCats=1024,
minSplitImprovement=1e-5,
histogramType=self._hc._jvm.hex.tree.SharedTreeModel.
SharedTreeParameters.HistogramType.valueOf("AUTO"),
r2Stopping=self._hc._jvm.Double.MAX_VALUE,
nbinsTopLevel=1 << 10,
buildTreeOneNode=False,
scoreTreeInterval=0,
sampleRate=1.0,
sampleRatePerClass=None,
colSampleRateChangePerLevel=1.0,
colSampleRatePerTree=1.0,
learnRate=0.1,
learnRateAnnealing=1.0,
colSampleRate=1.0,
maxAbsLeafnodePred=self._hc._jvm.Double.MAX_VALUE,
predNoiseBandwidth=0.0,
convertUnknownCategoricalLevelsToNa=False,
foldCol=None)
kwargs = get_input_kwargs(self)
self.setParams(**kwargs)
def setParams(self, keep=False, columns=[]):
kwargs = get_input_kwargs(self)
return self._set(**kwargs)
def __init__(self,
featuresCols=[],
labelCol="label",
allStringColumnsToCategorical=True,
columnsToCategorical=[],
splitRatio=1.0,
foldCol=None,
weightCol=None,
ignoredCols=[],
includeAlgos=None,
excludeAlgos=None,
projectName=None,
maxRuntimeSecs=3600.0,
stoppingRounds=3,
stoppingTolerance=0.001,
stoppingMetric="AUTO",
nfolds=5,
convertUnknownCategoricalLevelsToNa=True,
seed=-1,
sortMetric="AUTO",
balanceClasses=False,
classSamplingFactors=None,
maxAfterBalanceSize=5.0,
keepCrossValidationPredictions=True,
keepCrossValidationModels=True,
maxModels=0,
predictionCol="prediction",
detailedPredictionCol="detailed_prediction",
withDetailedPredictionCol=False,
convertInvalidNumbersToNa=False,
**deprecatedArgs):
super(H2OAutoML, self).__init__()
self._java_obj = self._new_java_obj("py_sparkling.ml.algos.H2OAutoML",
self.uid)
self._setDefault(featuresCols=[],
labelCol="label",
allStringColumnsToCategorical=True,
columnsToCategorical=[],
splitRatio=1.0,
foldCol=None,
weightCol=None,
ignoredCols=[],
includeAlgos=None,
excludeAlgos=None,
projectName=None,
maxRuntimeSecs=3600.0,
stoppingRounds=3,
stoppingTolerance=0.001,
stoppingMetric="AUTO",
nfolds=5,
convertUnknownCategoricalLevelsToNa=True,
seed=-1,
sortMetric=None,
balanceClasses=False,
classSamplingFactors=None,
maxAfterBalanceSize=5.0,
keepCrossValidationPredictions=True,
keepCrossValidationModels=True,
maxModels=0,
predictionCol="prediction",
detailedPredictionCol="detailed_prediction",
withDetailedPredictionCol=False,
convertInvalidNumbersToNa=False)
kwargs = get_input_kwargs(self)
self.setParams(**kwargs)
def setParams(self,
modelId=None,
splitRatio=1.0,
labelCol="label",
weightCol=None,
featuresCols=[],
allStringColumnsToCategorical=True,
columnsToCategorical=[],
nfolds=0,
keepCrossValidationPredictions=False,
keepCrossValidationFoldAssignment=False,
parallelizeCrossValidation=True,
seed=-1,
distribution="AUTO",
convertUnknownCategoricalLevelsToNa=False,
standardize=True,
family="gaussian",
link="family_default",
solver="AUTO",
tweedieVariancePower=0.0,
tweedieLinkPower=0.0,
alpha=None,
lambda_=None,
missingValuesHandling="MeanImputation",
prior=-1.0,
lambdaSearch=False,
nlambdas=-1,
nonNegative=False,
exactLambdas=False,
lambdaMinRatio=-1.0,
maxIterations=-1,
intercept=True,
betaEpsilon=1e-4,
objectiveEpsilon=-1.0,
gradientEpsilon=-1.0,
objReg=-1.0,
computePValues=False,
removeCollinearCols=False,
interactions=None,
interactionPairs=None,
earlyStopping=True,
foldCol=None,
**deprecatedArgs):
kwargs = get_input_kwargs(self)
if "distribution" in kwargs:
kwargs[
"distribution"] = self._hc._jvm.hex.genmodel.utils.DistributionFamily.valueOf(
kwargs["distribution"])
if "family" in kwargs:
kwargs[
"family"] = self._hc._jvm.hex.glm.GLMModel.GLMParameters.Family.valueOf(
kwargs["family"])
if "link" in kwargs:
kwargs[
"link"] = self._hc._jvm.hex.glm.GLMModel.GLMParameters.Link.valueOf(
kwargs["link"])
if "solver" in kwargs:
kwargs[
"solver"] = self._hc._jvm.hex.glm.GLMModel.GLMParameters.Solver.valueOf(
kwargs["solver"])
if "missingValuesHandling" in kwargs:
kwargs[
"missingValuesHandling"] = self._hc._jvm.hex.deeplearning.DeepLearningModel.DeepLearningParameters.MissingValuesHandling.valueOf(
kwargs["missingValuesHandling"])
# we need to convert double arguments manually to floats as if we assign integer to double, py4j thinks that
# the whole type is actually int and we get class cast exception
double_types = [
"splitRatio", "tweedieVariancePower", "tweedieLinkPower", "prior",
"lambdaMinRatio", "betaEpsilon", "objectiveEpsilon",
"gradientEpsilon", "objReg"
]
set_double_values(kwargs, double_types)
# We need to also map all doubles in the arrays
if "alpha" in kwargs:
kwargs["alpha"] = map(float, kwargs["alpha"])
if "lambda_" in kwargs:
kwargs["lambda_"] = map(float, kwargs["lambda_"])
return self._set(**kwargs)
def setParams(self,
modelId=None,
splitRatio=1.0,
labelCol="label",
weightCol=None,
featuresCols=[],
allStringColumnsToCategorical=True,
columnsToCategorical=[],
nfolds=0,
keepCrossValidationPredictions=False,
keepCrossValidationFoldAssignment=False,
parallelizeCrossValidation=True,
seed=-1,
distribution="AUTO",
ntrees=50,
maxDepth=5,
minRows=10.0,
nbins=20,
nbinsCats=1024,
minSplitImprovement=1e-5,
histogramType="AUTO",
r2Stopping=java_max_double_value,
nbinsTopLevel=1 << 10,
buildTreeOneNode=False,
scoreTreeInterval=0,
sampleRate=1.0,
sampleRatePerClass=None,
colSampleRateChangePerLevel=1.0,
colSampleRatePerTree=1.0,
learnRate=0.1,
learnRateAnnealing=1.0,
colSampleRate=1.0,
maxAbsLeafnodePred=java_max_double_value,
predNoiseBandwidth=0.0,
convertUnknownCategoricalLevelsToNa=False,
foldCol=None,
predictionCol="prediction",
detailedPredictionCol="detailed_prediction",
withDetailedPredictionCol=False,
convertInvalidNumbersToNa=False,
**deprecatedArgs):
kwargs = get_input_kwargs(self)
validateEnumValue(self._H2OAlgoCommonParams__getDistributionEnum(),
kwargs, "distribution")
validateEnumValue(self._H2OSharedTreeParams__getHistogramTypeEnum(),
kwargs, "histogramType")
# we need to convert double arguments manually to floats as if we assign integer to double, py4j thinks that
# the whole type is actually int and we get class cast exception
double_types = [
"minRows", "predNoiseBandwidth", "splitRatio", "learnRate",
"colSampleRate", "learnRateAnnealing", "maxAbsLeafnodePred"
"minSplitImprovement", "r2Stopping", "sampleRate",
"colSampleRateChangePerLevel", "colSampleRatePerTree"
]
set_double_values(kwargs, double_types)
# We need to also map all doubles in the arrays
arrayToDoubleArray("sampleRatePerClass", kwargs)
return self._set(**kwargs)
def setParams(self,
modelId=None,
splitRatio=1.0,
labelCol="label",
weightCol=None,
featuresCols=[],
allStringColumnsToCategorical=True,
columnsToCategorical=[],
nfolds=0,
keepCrossValidationPredictions=False,
keepCrossValidationFoldAssignment=False,
parallelizeCrossValidation=True,
seed=-1,
distribution="AUTO",
convertUnknownCategoricalLevelsToNa=False,
standardize=True,
family="gaussian",
link="family_default",
solver="AUTO",
tweedieVariancePower=0.0,
tweedieLinkPower=0.0,
alpha=None,
lambda_=None,
missingValuesHandling="MeanImputation",
prior=-1.0,
lambdaSearch=False,
nlambdas=-1,
nonNegative=False,
exactLambdas=False,
lambdaMinRatio=-1.0,
maxIterations=-1,
intercept=True,
betaEpsilon=1e-4,
objectiveEpsilon=-1.0,
gradientEpsilon=-1.0,
objReg=-1.0,
computePValues=False,
removeCollinearCols=False,
interactions=None,
interactionPairs=None,
earlyStopping=True,
foldCol=None,
predictionCol="prediction",
detailedPredictionCol="detailed_prediction",
withDetailedPredictionCol=False,
convertInvalidNumbersToNa=False,
**deprecatedArgs):
kwargs = get_input_kwargs(self)
validateEnumValue(self._H2OAlgoCommonParams__getDistributionEnum(),
kwargs, "distribution")
validateEnumValue(self._H2OGLMParams__getFamilyEnum(), kwargs,
"family")
validateEnumValue(self._H2OGLMParams__getLinkEnum(), kwargs, "link")
validateEnumValue(self._H2OGLMParams__getSolverEnum(), kwargs,
"solver")
validateEnumValue(self._H2OGLMParams__getMissingValuesHandlingEnum(),
kwargs, "missingValuesHandling")
# we need to convert double arguments manually to floats as if we assign integer to double, py4j thinks that
# the whole type is actually int and we get class cast exception
double_types = [
"splitRatio", "tweedieVariancePower", "tweedieLinkPower", "prior",
"lambdaMinRatio", "betaEpsilon", "objectiveEpsilon",
"gradientEpsilon", "objReg"
]
set_double_values(kwargs, double_types)
# We need to also map all doubles in the arrays
arrayToDoubleArray("alpha", kwargs)
arrayToDoubleArray("lambda_", kwargs)
return self._set(**kwargs)
def setParams(self,
modelId=None,
splitRatio=1.0,
labelCol="label",
weightCol=None,
featuresCols=[],
allStringColumnsToCategorical=True,
columnsToCategorical=[],
nfolds=0,
keepCrossValidationPredictions=False,
keepCrossValidationFoldAssignment=False,
parallelizeCrossValidation=True,
seed=-1,
distribution="AUTO",
convertUnknownCategoricalLevelsToNa=False,
quietMode=True,
ntrees=50,
nEstimators=0,
maxDepth=6,
minRows=1.0,
minChildWeight=1.0,
learnRate=0.3,
eta=0.3,
learnRateAnnealing=1.0,
sampleRate=1.0,
subsample=1.0,
colSampleRate=1.0,
colSampleByLevel=1.0,
colSampleRatePerTree=1.0,
colsampleBytree=1.0,
maxAbsLeafnodePred=0.0,
maxDeltaStep=0.0,
scoreTreeInterval=0,
initialScoreInterval=4000,
scoreInterval=4000,
minSplitImprovement=0.0,
gamma=0.0,
nthread=-1,
maxBins=256,
maxLeaves=0,
minSumHessianInLeaf=100.0,
minDataInLeaf=0.0,
treeMethod="auto",
growPolicy="depthwise",
booster="gbtree",
dmatrixType="auto",
regLambda=0.0,
regAlpha=0.0,
sampleType="uniform",
normalizeType="tree",
rateDrop=0.0,
oneDrop=False,
skipDrop=0.0,
gpuId=0,
backend="auto",
foldCol=None,
predictionCol="prediction",
detailedPredictionCol="detailed_prediction",
withDetailedPredictionCol=False,
convertInvalidNumbersToNa=False,
**deprecatedArgs):
kwargs = get_input_kwargs(self)
validateEnumValue(self._H2OAlgoCommonParams__getDistributionEnum(),
kwargs, "distribution")
validateEnumValue(self._H2OXGBoostParams__getTreeMethodEnum(), kwargs,
"treeMethod")
validateEnumValue(self._H2OXGBoostParams__getGrowPolicyEnum(), kwargs,
"growPolicy")
validateEnumValue(self._H2OXGBoostParams__getBoosterEnum(), kwargs,
"booster")
validateEnumValue(self._H2OXGBoostParams__getDmatrixTypeEnum(), kwargs,
"dmatrixType")
validateEnumValue(self._H2OXGBoostParams__getSampleTypeEnum(), kwargs,
"sampleType")
validateEnumValue(self._H2OXGBoostParams__getNormalizeTypeEnum(),
kwargs, "normalizeType")
validateEnumValue(self._H2OXGBoostParams__getBackendEnum(), kwargs,
"backend")
# we need to convert double arguments manually to floats as if we assign integer to double, py4j thinks that
# the whole type is actually int and we get class cast exception
double_types = [
"splitRatio", "minRows", "minChildWeight", "learnRate", "eta",
"learnRateAnnealing"
"sampleRate", "subsample", "colSampleRate", "colSampleByLevel",
"colSampleRatePerTree", "colsampleBytree", "maxAbsLeafnodePred",
"maxDeltaStep", "minSplitImprovement", "gamma",
"minSumHessianInLeaf", "minDataInLeaf", "regLambda", "regAlpha",
"rateDrop", "skipDrop"
]
set_double_values(kwargs, double_types)
return self._set(**kwargs)
def __init__(self,
modelId=None,
splitRatio=1.0,
labelCol="label",
weightCol=None,
featuresCols=[],
allStringColumnsToCategorical=True,
columnsToCategorical=[],
nfolds=0,
keepCrossValidationPredictions=False,
keepCrossValidationFoldAssignment=False,
parallelizeCrossValidation=True,
seed=-1,
distribution="AUTO",
convertUnknownCategoricalLevelsToNa=False,
quietMode=True,
ntrees=50,
nEstimators=0,
maxDepth=6,
minRows=1.0,
minChildWeight=1.0,
learnRate=0.3,
eta=0.3,
learnRateAnnealing=1.0,
sampleRate=1.0,
subsample=1.0,
colSampleRate=1.0,
colSampleByLevel=1.0,
colSampleRatePerTree=1.0,
colsampleBytree=1.0,
maxAbsLeafnodePred=0.0,
maxDeltaStep=0.0,
scoreTreeInterval=0,
initialScoreInterval=4000,
scoreInterval=4000,
minSplitImprovement=0.0,
gamma=0.0,
nthread=-1,
maxBins=256,
maxLeaves=0,
minSumHessianInLeaf=100.0,
minDataInLeaf=0.0,
treeMethod="auto",
growPolicy="depthwise",
booster="gbtree",
dmatrixType="auto",
regLambda=0.0,
regAlpha=0.0,
sampleType="uniform",
normalizeType="tree",
rateDrop=0.0,
oneDrop=False,
skipDrop=0.0,
gpuId=0,
backend="auto",
foldCol=None,
predictionCol="prediction",
detailedPredictionCol="detailed_prediction",
withDetailedPredictionCol=False,
convertInvalidNumbersToNa=False,
**deprecatedArgs):
super(H2OXGBoost, self).__init__()
self._java_obj = self._new_java_obj("py_sparkling.ml.algos.H2OXGBoost",
self.uid)
self._setDefault(modelId=None,
splitRatio=1.0,
labelCol="label",
weightCol=None,
featuresCols=[],
allStringColumnsToCategorical=True,
columnsToCategorical=[],
nfolds=0,
keepCrossValidationPredictions=False,
keepCrossValidationFoldAssignment=False,
parallelizeCrossValidation=True,
seed=-1,
distribution="AUTO",
convertUnknownCategoricalLevelsToNa=False,
quietMode=True,
ntrees=50,
nEstimators=0,
maxDepth=6,
minRows=1.0,
minChildWeight=1.0,
learnRate=0.3,
eta=0.3,
learnRateAnnealing=1.0,
sampleRate=1.0,
subsample=1.0,
colSampleRate=1.0,
colSampleByLevel=1.0,
colSampleRatePerTree=1.0,
colsampleBytree=1.0,
maxAbsLeafnodePred=0.0,
maxDeltaStep=0.0,
scoreTreeInterval=0,
initialScoreInterval=4000,
scoreInterval=4000,
minSplitImprovement=0.0,
gamma=0.0,
nthread=-1,
maxBins=256,
maxLeaves=0,
minSumHessianInLeaf=100.0,
minDataInLeaf=0.0,
treeMethod="auto",
growPolicy="depthwise",
booster="gbtree",
dmatrixType="auto",
regLambda=0.0,
regAlpha=0.0,
sampleType="uniform",
normalizeType="tree",
rateDrop=0.0,
oneDrop=False,
skipDrop=0.0,
gpuId=0,
backend="auto",
foldCol=None,
predictionCol="prediction",
detailedPredictionCol="detailed_prediction",
withDetailedPredictionCol=False,
convertInvalidNumbersToNa=False)
kwargs = get_input_kwargs(self)
self.setParams(**kwargs)
def setParams(self,
modelId=None,
splitRatio=1.0,
labelCol="label",
weightCol=None,
featuresCols=[],
allStringColumnsToCategorical=True,
columnsToCategorical=[],
nfolds=0,
keepCrossValidationPredictions=False,
keepCrossValidationFoldAssignment=False,
parallelizeCrossValidation=True,
seed=-1,
distribution="AUTO",
convertUnknownCategoricalLevelsToNa=False,
quietMode=True,
missingValuesHandling=None,
ntrees=50,
nEstimators=0,
maxDepth=6,
minRows=1.0,
minChildWeight=1.0,
learnRate=0.3,
eta=0.3,
learnRateAnnealing=1.0,
sampleRate=1.0,
subsample=1.0,
colSampleRate=1.0,
colSampleByLevel=1.0,
colSampleRatePerTree=1.0,
colsampleBytree=1.0,
maxAbsLeafnodePred=0.0,
maxDeltaStep=0.0,
scoreTreeInterval=0,
initialScoreInterval=4000,
scoreInterval=4000,
minSplitImprovement=0.0,
gamma=0.0,
nthread=-1,
maxBins=256,
maxLeaves=0,
minSumHessianInLeaf=100.0,
minDataInLeaf=0.0,
treeMethod="auto",
growPolicy="depthwise",
booster="gbtree",
dmatrixType="auto",
regLambda=0.0,
regAlpha=0.0,
sampleType="uniform",
normalizeType="tree",
rateDrop=0.0,
oneDrop=False,
skipDrop=0.0,
gpuId=0,
backend="auto",
foldCol=None,
**deprecatedArgs):
kwargs = get_input_kwargs(self)
if "distribution" in kwargs:
kwargs[
"distribution"] = self._hc._jvm.hex.genmodel.utils.DistributionFamily.valueOf(
kwargs["distribution"])
if "treeMethod" in kwargs:
kwargs[
"treeMethod"] = self._hc._jvm.hex.tree.xgboost.XGBoostModel.XGBoostParameters.TreeMethod.valueOf(
kwargs["treeMethod"])
if "growPolicy" in kwargs:
kwargs[
"growPolicy"] = self._hc._jvm.hex.tree.xgboost.XGBoostModel.XGBoostParameters.GrowPolicy.valueOf(
kwargs["growPolicy"])
if "booster" in kwargs:
kwargs[
"booster"] = self._hc._jvm.hex.tree.xgboost.XGBoostModel.XGBoostParameters.Booster.valueOf(
kwargs["booster"])
if "dmatrixType" in kwargs:
kwargs[
"dmatrixType"] = self._hc._jvm.hex.tree.xgboost.XGBoostModel.XGBoostParameters.DMatrixType.valueOf(
kwargs["dmatrixType"])
if "sampleType" in kwargs:
kwargs[
"sampleType"] = self._hc._jvm.hex.tree.xgboost.XGBoostModel.XGBoostParameters.DartSampleType.valueOf(
kwargs["sampleType"])
if "normalizeType" in kwargs:
kwargs[
"normalizeType"] = self._hc._jvm.hex.tree.xgboost.XGBoostModel.XGBoostParameters.DartNormalizeType.valueOf(
kwargs["normalizeType"])
if "backend" in kwargs:
kwargs[
"backend"] = self._hc._jvm.hex.tree.xgboost.XGBoostModel.XGBoostParameters.Backend.valueOf(
kwargs["backend"])
# we need to convert double arguments manually to floats as if we assign integer to double, py4j thinks that
# the whole type is actually int and we get class cast exception
double_types = [
"splitRatio", "minRows", "minChildWeight", "learnRate", "eta",
"learnRateAnnealing"
"sampleRate", "subsample", "colSampleRate", "colSampleByLevel",
"colSampleRatePerTree", "colsampleBytree", "maxAbsLeafnodePred",
"maxDeltaStep", "minSplitImprovement", "gamma",
"minSumHessianInLeaf", "minDataInLeaf", "regLambda", "regAlpha",
"rateDrop", "skipDrop"
]
set_double_values(kwargs, double_types)
return self._set(**kwargs)
def __init__(self,
modelId=None,
splitRatio=1.0,
labelCol="label",
weightCol=None,
featuresCols=[],
allStringColumnsToCategorical=True,
columnsToCategorical=[],
nfolds=0,
keepCrossValidationPredictions=False,
keepCrossValidationFoldAssignment=False,
parallelizeCrossValidation=True,
seed=-1,
distribution="AUTO",
convertUnknownCategoricalLevelsToNa=False,
standardize=True,
family="gaussian",
link="family_default",
solver="AUTO",
tweedieVariancePower=0.0,
tweedieLinkPower=0.0,
alpha=None,
lambda_=None,
missingValuesHandling="MeanImputation",
prior=-1.0,
lambdaSearch=False,
nlambdas=-1,
nonNegative=False,
exactLambdas=False,
lambdaMinRatio=-1.0,
maxIterations=-1,
intercept=True,
betaEpsilon=1e-4,
objectiveEpsilon=-1.0,
gradientEpsilon=-1.0,
objReg=-1.0,
computePValues=False,
removeCollinearCols=False,
interactions=None,
interactionPairs=None,
earlyStopping=True,
foldCol=None,
**deprecatedArgs):
super(H2OGLM, self).__init__()
self._hc = H2OContext.getOrCreate(SparkSession.builder.getOrCreate(),
verbose=False)
self._java_obj = self._new_java_obj("py_sparkling.ml.algos.H2OGLM",
self.uid)
self._setDefault(
modelId=None,
splitRatio=1.0,
labelCol="label",
weightCol=None,
featuresCols=[],
allStringColumnsToCategorical=True,
columnsToCategorical=[],
nfolds=0,
keepCrossValidationPredictions=False,
keepCrossValidationFoldAssignment=False,
parallelizeCrossValidation=True,
seed=-1,
distribution=self._hc._jvm.hex.genmodel.utils.DistributionFamily.
valueOf("AUTO"),
convertUnknownCategoricalLevelsToNa=False,
standardize=True,
family=self._hc._jvm.hex.glm.GLMModel.GLMParameters.Family.valueOf(
"gaussian"),
link=self._hc._jvm.hex.glm.GLMModel.GLMParameters.Link.valueOf(
"family_default"),
solver=self._hc._jvm.hex.glm.GLMModel.GLMParameters.Solver.valueOf(
"AUTO"),
tweedieVariancePower=0.0,
tweedieLinkPower=0.0,
alpha=None,
lambda_=None,
missingValuesHandling=self._hc._jvm.hex.deeplearning.
DeepLearningModel.DeepLearningParameters.MissingValuesHandling.
valueOf("MeanImputation"),
prior=-1.0,
lambdaSearch=False,
nlambdas=-1,
nonNegative=False,
exactLambdas=False,
lambdaMinRatio=-1.0,
maxIterations=-1,
intercept=True,
betaEpsilon=1e-4,
objectiveEpsilon=-1.0,
gradientEpsilon=-1.0,
objReg=-1.0,
computePValues=False,
removeCollinearCols=False,
interactions=None,
interactionPairs=None,
earlyStopping=True,
foldCol=None)
kwargs = get_input_kwargs(self)
self.setParams(**kwargs)
def setParams(self,
modelId=None,
splitRatio=1.0,
labelCol="label",
weightCol=None,
featuresCols=[],
allStringColumnsToCategorical=True,
columnsToCategorical=[],
nfolds=0,
keepCrossValidationPredictions=False,
keepCrossValidationFoldAssignment=False,
parallelizeCrossValidation=True,
seed=-1,
distribution="AUTO",
ntrees=50,
maxDepth=5,
minRows=10.0,
nbins=20,
nbinsCats=1024,
minSplitImprovement=1e-5,
histogramType="AUTO",
r2Stopping=java_max_double_value,
nbinsTopLevel=1 << 10,
buildTreeOneNode=False,
scoreTreeInterval=0,
sampleRate=1.0,
sampleRatePerClass=None,
colSampleRateChangePerLevel=1.0,
colSampleRatePerTree=1.0,
learnRate=0.1,
learnRateAnnealing=1.0,
colSampleRate=1.0,
maxAbsLeafnodePred=java_max_double_value,
predNoiseBandwidth=0.0,
convertUnknownCategoricalLevelsToNa=False,
foldCol=None,
**deprecatedArgs):
kwargs = get_input_kwargs(self)
if "distribution" in kwargs:
kwargs[
"distribution"] = self._hc._jvm.hex.genmodel.utils.DistributionFamily.valueOf(
kwargs["distribution"])
if "histogramType" in kwargs:
kwargs[
"histogramType"] = self._hc._jvm.hex.tree.SharedTreeModel.SharedTreeParameters.HistogramType.valueOf(
kwargs["histogramType"])
# we need to convert double arguments manually to floats as if we assign integer to double, py4j thinks that
# the whole type is actually int and we get class cast exception
double_types = [
"minRows", "predNoiseBandwidth", "splitRatio", "learnRate",
"colSampleRate", "learnRateAnnealing", "maxAbsLeafnodePred"
"minSplitImprovement", "r2Stopping", "sampleRate",
"colSampleRateChangePerLevel", "colSampleRatePerTree"
]
set_double_values(kwargs, double_types)
# We need to also map all doubles in the arrays
if "sampleRatePerClass" in kwargs:
kwargs["sampleRatePerClass"] = map(float,
kwargs["sampleRatePerClass"])
return self._set(**kwargs)
def __init__(self,
modelId=None,
splitRatio=1.0,
labelCol="label",
weightCol=None,
featuresCols=[],
allStringColumnsToCategorical=True,
columnsToCategorical=[],
nfolds=0,
keepCrossValidationPredictions=False,
keepCrossValidationFoldAssignment=False,
parallelizeCrossValidation=True,
seed=-1,
distribution="AUTO",
convertUnknownCategoricalLevelsToNa=False,
standardize=True,
family="gaussian",
link="family_default",
solver="AUTO",
tweedieVariancePower=0.0,
tweedieLinkPower=0.0,
alpha=None,
lambda_=None,
missingValuesHandling="MeanImputation",
prior=-1.0,
lambdaSearch=False,
nlambdas=-1,
nonNegative=False,
exactLambdas=False,
lambdaMinRatio=-1.0,
maxIterations=-1,
intercept=True,
betaEpsilon=1e-4,
objectiveEpsilon=-1.0,
gradientEpsilon=-1.0,
objReg=-1.0,
computePValues=False,
removeCollinearCols=False,
interactions=None,
interactionPairs=None,
earlyStopping=True,
foldCol=None,
predictionCol="prediction",
detailedPredictionCol="detailed_prediction",
withDetailedPredictionCol=False,
convertInvalidNumbersToNa=False,
**deprecatedArgs):
super(H2OGLM, self).__init__()
self._java_obj = self._new_java_obj("py_sparkling.ml.algos.H2OGLM",
self.uid)
self._setDefault(modelId=None,
splitRatio=1.0,
labelCol="label",
weightCol=None,
featuresCols=[],
allStringColumnsToCategorical=True,
columnsToCategorical=[],
nfolds=0,
keepCrossValidationPredictions=False,
keepCrossValidationFoldAssignment=False,
parallelizeCrossValidation=True,
seed=-1,
distribution="AUTO",
convertUnknownCategoricalLevelsToNa=False,
standardize=True,
family="gaussian",
link="family_default",
solver="AUTO",
tweedieVariancePower=0.0,
tweedieLinkPower=0.0,
alpha=None,
lambda_=None,
missingValuesHandling="MeanImputation",
prior=-1.0,
lambdaSearch=False,
nlambdas=-1,
nonNegative=False,
exactLambdas=False,
lambdaMinRatio=-1.0,
maxIterations=-1,
intercept=True,
betaEpsilon=1e-4,
objectiveEpsilon=-1.0,
gradientEpsilon=-1.0,
objReg=-1.0,
computePValues=False,
removeCollinearCols=False,
interactions=None,
interactionPairs=None,
earlyStopping=True,
foldCol=None,
predictionCol="prediction",
detailedPredictionCol="detailed_prediction",
withDetailedPredictionCol=False,
convertInvalidNumbersToNa=False)
kwargs = get_input_kwargs(self)
self.setParams(**kwargs)
def __init__(self,
modelId=None,
splitRatio=1.0,
labelCol="label",
weightCol=None,
featuresCols=[],
allStringColumnsToCategorical=True,
columnsToCategorical=[],
nfolds=0,
keepCrossValidationPredictions=False,
keepCrossValidationFoldAssignment=False,
parallelizeCrossValidation=True,
seed=-1,
distribution="AUTO",
ntrees=50,
maxDepth=5,
minRows=10.0,
nbins=20,
nbinsCats=1024,
minSplitImprovement=1e-5,
histogramType="AUTO",
r2Stopping=java_max_double_value,
nbinsTopLevel=1 << 10,
buildTreeOneNode=False,
scoreTreeInterval=0,
sampleRate=1.0,
sampleRatePerClass=None,
colSampleRateChangePerLevel=1.0,
colSampleRatePerTree=1.0,
learnRate=0.1,
learnRateAnnealing=1.0,
colSampleRate=1.0,
maxAbsLeafnodePred=java_max_double_value,
predNoiseBandwidth=0.0,
convertUnknownCategoricalLevelsToNa=False,
foldCol=None,
predictionCol="prediction",
detailedPredictionCol="detailed_prediction",
withDetailedPredictionCol=False,
convertInvalidNumbersToNa=False,
**deprecatedArgs):
super(H2OGBM, self).__init__()
self._java_obj = self._new_java_obj("py_sparkling.ml.algos.H2OGBM",
self.uid)
self._setDefault(modelId=None,
splitRatio=1.0,
labelCol="label",
weightCol=None,
featuresCols=[],
allStringColumnsToCategorical=True,
columnsToCategorical=[],
nfolds=0,
keepCrossValidationPredictions=False,
keepCrossValidationFoldAssignment=False,
parallelizeCrossValidation=True,
seed=-1,
distribution="AUTO",
ntrees=50,
maxDepth=5,
minRows=10.0,
nbins=20,
nbinsCats=1024,
minSplitImprovement=1e-5,
histogramType="AUTO",
r2Stopping=_jvm().Double.MAX_VALUE,
nbinsTopLevel=1 << 10,
buildTreeOneNode=False,
scoreTreeInterval=0,
sampleRate=1.0,
sampleRatePerClass=None,
colSampleRateChangePerLevel=1.0,
colSampleRatePerTree=1.0,
learnRate=0.1,
learnRateAnnealing=1.0,
colSampleRate=1.0,
maxAbsLeafnodePred=_jvm().Double.MAX_VALUE,
predNoiseBandwidth=0.0,
convertUnknownCategoricalLevelsToNa=False,
foldCol=None,
predictionCol="prediction",
detailedPredictionCol="detailed_prediction",
withDetailedPredictionCol=False,
convertInvalidNumbersToNa=False)
kwargs = get_input_kwargs(self)
self.setParams(**kwargs)