python/pyspark/ml/tuning.py

#
# Licensed to the Apache Software Foundation (ASF) under one or more
# contributor license agreements.  See the NOTICE file distributed with
# this work for additional information regarding copyright ownership.
# The ASF licenses this file to You under the Apache License, Version 2.0
# (the "License"); you may not use this file except in compliance with
# the License.  You may obtain a copy of the License at
#
#    http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
#

import itertools
import numpy as np

from pyspark import since, keyword_only
from pyspark.ml import Estimator, Model
from pyspark.ml.param import Params, Param, TypeConverters
from pyspark.ml.param.shared import HasSeed
from pyspark.sql.functions import rand

__all__ = ['ParamGridBuilder', 'CrossValidator', 'CrossValidatorModel', 'TrainValidationSplit',
           'TrainValidationSplitModel']


class ParamGridBuilder(object):
    r"""
    Builder for a param grid used in grid search-based model selection.

    >>> from pyspark.ml.classification import LogisticRegression
    >>> lr = LogisticRegression()
    >>> output = ParamGridBuilder() \
    ...     .baseOn({lr.labelCol: 'l'}) \
    ...     .baseOn([lr.predictionCol, 'p']) \
    ...     .addGrid(lr.regParam, [1.0, 2.0]) \
    ...     .addGrid(lr.maxIter, [1, 5]) \
    ...     .build()
    >>> expected = [
    ...     {lr.regParam: 1.0, lr.maxIter: 1, lr.labelCol: 'l', lr.predictionCol: 'p'},
    ...     {lr.regParam: 2.0, lr.maxIter: 1, lr.labelCol: 'l', lr.predictionCol: 'p'},
    ...     {lr.regParam: 1.0, lr.maxIter: 5, lr.labelCol: 'l', lr.predictionCol: 'p'},
    ...     {lr.regParam: 2.0, lr.maxIter: 5, lr.labelCol: 'l', lr.predictionCol: 'p'}]
    >>> len(output) == len(expected)
    True
    >>> all([m in expected for m in output])
    True

    .. versionadded:: 1.4.0
    """

    def __init__(self):
        self._param_grid = {}

    @since("1.4.0")
    def addGrid(self, param, values):
        """
        Sets the given parameters in this grid to fixed values.
        """
        self._param_grid[param] = values

        return self

    @since("1.4.0")
    def baseOn(self, *args):
        """
        Sets the given parameters in this grid to fixed values.
        Accepts either a parameter dictionary or a list of (parameter, value) pairs.
        """
        if isinstance(args[0], dict):
            self.baseOn(*args[0].items())
        else:
            for (param, value) in args:
                self.addGrid(param, [value])

        return self

    @since("1.4.0")
    def build(self):
        """
        Builds and returns all combinations of parameters specified
        by the param grid.
        """
        keys = self._param_grid.keys()
        grid_values = self._param_grid.values()
        return [dict(zip(keys, prod)) for prod in itertools.product(*grid_values)]


class ValidatorParams(HasSeed):
    """
    Common params for TrainValidationSplit and CrossValidator.
    """

    estimator = Param(Params._dummy(), "estimator", "estimator to be cross-validated")
    estimatorParamMaps = Param(Params._dummy(), "estimatorParamMaps", "estimator param maps")
    evaluator = Param(
        Params._dummy(), "evaluator",
        "evaluator used to select hyper-parameters that maximize the validator metric")

    def setEstimator(self, value):
        """
        Sets the value of :py:attr:`estimator`.
        """
        return self._set(estimator=value)

    def getEstimator(self):
        """
        Gets the value of estimator or its default value.
        """
        return self.getOrDefault(self.estimator)

    def setEstimatorParamMaps(self, value):
        """
        Sets the value of :py:attr:`estimatorParamMaps`.
        """
        return self._set(estimatorParamMaps=value)

    def getEstimatorParamMaps(self):
        """
        Gets the value of estimatorParamMaps or its default value.
        """
        return self.getOrDefault(self.estimatorParamMaps)

    def setEvaluator(self, value):
        """
        Sets the value of :py:attr:`evaluator`.
        """
        return self._set(evaluator=value)

    def getEvaluator(self):
        """
        Gets the value of evaluator or its default value.
        """
        return self.getOrDefault(self.evaluator)


class CrossValidator(Estimator, ValidatorParams):
    """

    K-fold cross validation performs model selection by splitting the dataset into a set of
    non-overlapping randomly partitioned folds which are used as separate training and test datasets
    e.g., with k=3 folds, K-fold cross validation will generate 3 (training, test) dataset pairs,
    each of which uses 2/3 of the data for training and 1/3 for testing. Each fold is used as the
    test set exactly once.


    >>> from pyspark.ml.classification import LogisticRegression
    >>> from pyspark.ml.evaluation import BinaryClassificationEvaluator
    >>> from pyspark.ml.linalg import Vectors
    >>> dataset = spark.createDataFrame(
    ...     [(Vectors.dense([0.0]), 0.0),
    ...      (Vectors.dense([0.4]), 1.0),
    ...      (Vectors.dense([0.5]), 0.0),
    ...      (Vectors.dense([0.6]), 1.0),
    ...      (Vectors.dense([1.0]), 1.0)] * 10,
    ...     ["features", "label"])
    >>> lr = LogisticRegression()
    >>> grid = ParamGridBuilder().addGrid(lr.maxIter, [0, 1]).build()
    >>> evaluator = BinaryClassificationEvaluator()
    >>> cv = CrossValidator(estimator=lr, estimatorParamMaps=grid, evaluator=evaluator)
    >>> cvModel = cv.fit(dataset)
    >>> cvModel.avgMetrics[0]
    0.5
    >>> evaluator.evaluate(cvModel.transform(dataset))
    0.8333...

    .. versionadded:: 1.4.0
    """

    numFolds = Param(Params._dummy(), "numFolds", "number of folds for cross validation",
                     typeConverter=TypeConverters.toInt)

    @keyword_only
    def __init__(self, estimator=None, estimatorParamMaps=None, evaluator=None, numFolds=3,
                 seed=None):
        """
        __init__(self, estimator=None, estimatorParamMaps=None, evaluator=None, numFolds=3,\
                 seed=None)
        """
        super(CrossValidator, self).__init__()
        self._setDefault(numFolds=3)
        kwargs = self._input_kwargs
        self._set(**kwargs)

    @keyword_only
    @since("1.4.0")
    def setParams(self, estimator=None, estimatorParamMaps=None, evaluator=None, numFolds=3,
                  seed=None):
        """
        setParams(self, estimator=None, estimatorParamMaps=None, evaluator=None, numFolds=3,\
                  seed=None):
        Sets params for cross validator.
        """
        kwargs = self._input_kwargs
        return self._set(**kwargs)

    @since("1.4.0")
    def setNumFolds(self, value):
        """
        Sets the value of :py:attr:`numFolds`.
        """
        return self._set(numFolds=value)

    @since("1.4.0")
    def getNumFolds(self):
        """
        Gets the value of numFolds or its default value.
        """
        return self.getOrDefault(self.numFolds)

    def _fit(self, dataset):
        est = self.getOrDefault(self.estimator)
        epm = self.getOrDefault(self.estimatorParamMaps)
        numModels = len(epm)
        eva = self.getOrDefault(self.evaluator)
        nFolds = self.getOrDefault(self.numFolds)
        seed = self.getOrDefault(self.seed)
        h = 1.0 / nFolds
        randCol = self.uid + "_rand"
        df = dataset.select("*", rand(seed).alias(randCol))
        metrics = [0.0] * numModels
        for i in range(nFolds):
            validateLB = i * h
            validateUB = (i + 1) * h
            condition = (df[randCol] >= validateLB) & (df[randCol] < validateUB)
            validation = df.filter(condition)
            train = df.filter(~condition)
            models = est.fit(train, epm)
            for j in range(numModels):
                model = models[j]
                # TODO: duplicate evaluator to take extra params from input
                metric = eva.evaluate(model.transform(validation, epm[j]))
                metrics[j] += metric/nFolds

        if eva.isLargerBetter():
            bestIndex = np.argmax(metrics)
        else:
            bestIndex = np.argmin(metrics)
        bestModel = est.fit(dataset, epm[bestIndex])
        return self._copyValues(CrossValidatorModel(bestModel, metrics))

    @since("1.4.0")
    def copy(self, extra=None):
        """
        Creates a copy of this instance with a randomly generated uid
        and some extra params. This copies creates a deep copy of
        the embedded paramMap, and copies the embedded and extra parameters over.

        :param extra: Extra parameters to copy to the new instance
        :return: Copy of this instance
        """
        if extra is None:
            extra = dict()
        newCV = Params.copy(self, extra)
        if self.isSet(self.estimator):
            newCV.setEstimator(self.getEstimator().copy(extra))
        # estimatorParamMaps remain the same
        if self.isSet(self.evaluator):
            newCV.setEvaluator(self.getEvaluator().copy(extra))
        return newCV


class CrossValidatorModel(Model, ValidatorParams):
    """

    CrossValidatorModel contains the model with the highest average cross-validation
    metric across folds and uses this model to transform input data. CrossValidatorModel
    also tracks the metrics for each param map evaluated.

    .. versionadded:: 1.4.0
    """

    def __init__(self, bestModel, avgMetrics=[]):
        super(CrossValidatorModel, self).__init__()
        #: best model from cross validation
        self.bestModel = bestModel
        #: Average cross-validation metrics for each paramMap in
        #: CrossValidator.estimatorParamMaps, in the corresponding order.
        self.avgMetrics = avgMetrics

    def _transform(self, dataset):
        return self.bestModel.transform(dataset)

    @since("1.4.0")
    def copy(self, extra=None):
        """
        Creates a copy of this instance with a randomly generated uid
        and some extra params. This copies the underlying bestModel,
        creates a deep copy of the embedded paramMap, and
        copies the embedded and extra parameters over.

        :param extra: Extra parameters to copy to the new instance
        :return: Copy of this instance
        """
        if extra is None:
            extra = dict()
        bestModel = self.bestModel.copy(extra)
        avgMetrics = self.avgMetrics
        return CrossValidatorModel(bestModel, avgMetrics)


class TrainValidationSplit(Estimator, ValidatorParams):
    """
    .. note:: Experimental

    Validation for hyper-parameter tuning. Randomly splits the input dataset into train and
    validation sets, and uses evaluation metric on the validation set to select the best model.
    Similar to :class:`CrossValidator`, but only splits the set once.

    >>> from pyspark.ml.classification import LogisticRegression
    >>> from pyspark.ml.evaluation import BinaryClassificationEvaluator
    >>> from pyspark.ml.linalg import Vectors
    >>> dataset = spark.createDataFrame(
    ...     [(Vectors.dense([0.0]), 0.0),
    ...      (Vectors.dense([0.4]), 1.0),
    ...      (Vectors.dense([0.5]), 0.0),
    ...      (Vectors.dense([0.6]), 1.0),
    ...      (Vectors.dense([1.0]), 1.0)] * 10,
    ...     ["features", "label"])
    >>> lr = LogisticRegression()
    >>> grid = ParamGridBuilder().addGrid(lr.maxIter, [0, 1]).build()
    >>> evaluator = BinaryClassificationEvaluator()
    >>> tvs = TrainValidationSplit(estimator=lr, estimatorParamMaps=grid, evaluator=evaluator)
    >>> tvsModel = tvs.fit(dataset)
    >>> evaluator.evaluate(tvsModel.transform(dataset))
    0.8333...

    .. versionadded:: 2.0.0
    """

    trainRatio = Param(Params._dummy(), "trainRatio", "Param for ratio between train and\
     validation data. Must be between 0 and 1.", typeConverter=TypeConverters.toFloat)

    @keyword_only
    def __init__(self, estimator=None, estimatorParamMaps=None, evaluator=None, trainRatio=0.75,
                 seed=None):
        """
        __init__(self, estimator=None, estimatorParamMaps=None, evaluator=None, trainRatio=0.75,\
                 seed=None)
        """
        super(TrainValidationSplit, self).__init__()
        self._setDefault(trainRatio=0.75)
        kwargs = self._input_kwargs
        self._set(**kwargs)

    @since("2.0.0")
    @keyword_only
    def setParams(self, estimator=None, estimatorParamMaps=None, evaluator=None, trainRatio=0.75,
                  seed=None):
        """
        setParams(self, estimator=None, estimatorParamMaps=None, evaluator=None, trainRatio=0.75,\
                  seed=None):
        Sets params for the train validation split.
        """
        kwargs = self._input_kwargs
        return self._set(**kwargs)

    @since("2.0.0")
    def setTrainRatio(self, value):
        """
        Sets the value of :py:attr:`trainRatio`.
        """
        return self._set(trainRatio=value)

    @since("2.0.0")
    def getTrainRatio(self):
        """
        Gets the value of trainRatio or its default value.
        """
        return self.getOrDefault(self.trainRatio)

    def _fit(self, dataset):
        est = self.getOrDefault(self.estimator)
        epm = self.getOrDefault(self.estimatorParamMaps)
        numModels = len(epm)
        eva = self.getOrDefault(self.evaluator)
        tRatio = self.getOrDefault(self.trainRatio)
        seed = self.getOrDefault(self.seed)
        randCol = self.uid + "_rand"
        df = dataset.select("*", rand(seed).alias(randCol))
        metrics = [0.0] * numModels
        condition = (df[randCol] >= tRatio)
        validation = df.filter(condition)
        train = df.filter(~condition)
        models = est.fit(train, epm)
        for j in range(numModels):
            model = models[j]
            metric = eva.evaluate(model.transform(validation, epm[j]))
            metrics[j] += metric
        if eva.isLargerBetter():
            bestIndex = np.argmax(metrics)
        else:
            bestIndex = np.argmin(metrics)
        bestModel = est.fit(dataset, epm[bestIndex])
        return self._copyValues(TrainValidationSplitModel(bestModel, metrics))

    @since("2.0.0")
    def copy(self, extra=None):
        """
        Creates a copy of this instance with a randomly generated uid
        and some extra params. This copies creates a deep copy of
        the embedded paramMap, and copies the embedded and extra parameters over.

        :param extra: Extra parameters to copy to the new instance
        :return: Copy of this instance
        """
        if extra is None:
            extra = dict()
        newTVS = Params.copy(self, extra)
        if self.isSet(self.estimator):
            newTVS.setEstimator(self.getEstimator().copy(extra))
        # estimatorParamMaps remain the same
        if self.isSet(self.evaluator):
            newTVS.setEvaluator(self.getEvaluator().copy(extra))
        return newTVS


class TrainValidationSplitModel(Model, ValidatorParams):
    """
    .. note:: Experimental

    Model from train validation split.

    .. versionadded:: 2.0.0
    """

    def __init__(self, bestModel, validationMetrics=[]):
        super(TrainValidationSplitModel, self).__init__()
        #: best model from cross validation
        self.bestModel = bestModel
        #: evaluated validation metrics
        self.validationMetrics = validationMetrics

    def _transform(self, dataset):
        return self.bestModel.transform(dataset)

    @since("2.0.0")
    def copy(self, extra=None):
        """
        Creates a copy of this instance with a randomly generated uid
        and some extra params. This copies the underlying bestModel,
        creates a deep copy of the embedded paramMap, and
        copies the embedded and extra parameters over.
        And, this creates a shallow copy of the validationMetrics.

        :param extra: Extra parameters to copy to the new instance
        :return: Copy of this instance
        """
        if extra is None:
            extra = dict()
        bestModel = self.bestModel.copy(extra)
        validationMetrics = list(self.validationMetrics)
        return TrainValidationSplitModel(bestModel, validationMetrics)


if __name__ == "__main__":
    import doctest

    from pyspark.sql import SparkSession
    globs = globals().copy()

    # The small batch size here ensures that we see multiple batches,
    # even in these small test examples:
    spark = SparkSession.builder\
        .master("local[2]")\
        .appName("ml.tuning tests")\
        .getOrCreate()
    sc = spark.sparkContext
    globs['sc'] = sc
    globs['spark'] = spark
    (failure_count, test_count) = doctest.testmod(globs=globs, optionflags=doctest.ELLIPSIS)
    spark.stop()
    if failure_count:
        exit(-1)