[GitHub] spark pull request #17419: [SPARK-19634][ML] Multivariate summarizer - dataf...
Github user thunterdb closed the pull request at: https://github.com/apache/spark/pull/17419 --- If your project is set up for it, you can reply to this email and have your reply appear on GitHub as well. If your project does not have this feature enabled and wishes so, or if the feature is enabled but not working, please contact infrastructure at infrastruct...@apache.org or file a JIRA ticket with INFRA. --- - To unsubscribe, e-mail: reviews-unsubscr...@spark.apache.org For additional commands, e-mail: reviews-h...@spark.apache.org
[GitHub] spark pull request #17419: [SPARK-19634][ML] Multivariate summarizer - dataf...
Github user yanboliang commented on a diff in the pull request:
https://github.com/apache/spark/pull/17419#discussion_r130517322
--- Diff: mllib/src/main/scala/org/apache/spark/ml/stat/Summarizer.scala ---
@@ -0,0 +1,799 @@
+/*
+ * 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.
+ */
+
+package org.apache.spark.ml.stat
+
+import breeze.{linalg => la}
+import breeze.linalg.{Vector => BV}
+import breeze.numerics
+
+import org.apache.spark.SparkException
+import org.apache.spark.annotation.Since
+import org.apache.spark.internal.Logging
+import org.apache.spark.ml.linalg.{DenseVector, SparseVector, Vector,
Vectors, VectorUDT}
+import org.apache.spark.sql.Column
+import org.apache.spark.sql.catalyst.InternalRow
+import org.apache.spark.sql.catalyst.expressions.{Expression,
UnsafeArrayData, UnsafeProjection, UnsafeRow}
+import
org.apache.spark.sql.catalyst.expressions.aggregate.{AggregateExpression,
Complete, TypedImperativeAggregate}
+import org.apache.spark.sql.types._
+
+
+/**
+ * A builder object that provides summary statistics about a given column.
+ *
+ * Users should not directly create such builders, but instead use one of
the methods in
+ * [[Summarizer]].
+ */
+@Since("2.2.0")
+abstract class SummaryBuilder {
+ /**
+ * Returns an aggregate object that contains the summary of the column
with the requested metrics.
+ * @param column a column that contains Vector object.
+ * @return an aggregate column that contains the statistics. The exact
content of this
+ * structure is determined during the creation of the builder.
+ */
+ @Since("2.2.0")
+ def summary(column: Column): Column
+}
+
+/**
+ * Tools for vectorized statistics on MLlib Vectors.
+ *
+ * The methods in this package provide various statistics for Vectors
contained inside DataFrames.
+ *
+ * This class lets users pick the statistics they would like to extract
for a given column. Here is
+ * an example in Scala:
+ * {{{
+ * val dataframe = ... // Some dataframe containing a feature column
+ * val allStats = dataframe.select(Summarizer.metrics("min",
"max").summary($"features"))
+ * val Row(min_, max_) = allStats.first()
+ * }}}
+ *
+ * If one wants to get a single metric, shortcuts are also available:
+ * {{{
+ * val meanDF = dataframe.select(Summarizer.mean($"features"))
+ * val Row(mean_) = meanDF.first()
+ * }}}
+ */
+@Since("2.2.0")
+object Summarizer extends Logging {
+
+ import SummaryBuilderImpl._
+
+ /**
+ * Given a list of metrics, provides a builder that it turns computes
metrics from a column.
+ *
+ * See the documentation of [[Summarizer]] for an example.
+ *
+ * The following metrics are accepted (case sensitive):
+ * - mean: a vector that contains the coefficient-wise mean.
+ * - variance: a vector tha contains the coefficient-wise variance.
+ * - count: the count of all vectors seen.
+ * - numNonzeros: a vector with the number of non-zeros for each
coefficients
+ * - max: the maximum for each coefficient.
+ * - min: the minimum for each coefficient.
+ * - normL2: the Euclidian norm for each coefficient.
+ * - normL1: the L1 norm of each coefficient (sum of the absolute
values).
+ * @param firstMetric the metric being provided
+ * @param metrics additional metrics that can be provided.
+ * @return a builder.
+ * @throws IllegalArgumentException if one of the metric names is not
understood.
+ */
+ @Since("2.2.0")
+ def metrics(firstMetric: String, metrics: String*): SummaryBuilder = {
+val (typedMetrics, computeMetrics) =
getRelevantMetrics(Seq(firstMetric) ++ metrics)
+new SummaryBuilderImpl(typedMetrics, computeMetrics)
+ }
+
+ def mean(col: Column): Column = getSingleMetric(col, "mean")
+
+ def variance(col: Column): Column = getSingleMetric(col, "variance")
[GitHub] spark pull request #17419: [SPARK-19634][ML] Multivariate summarizer - dataf...
Github user yanboliang commented on a diff in the pull request:
https://github.com/apache/spark/pull/17419#discussion_r130515498
--- Diff: mllib/src/main/scala/org/apache/spark/ml/stat/Summarizer.scala ---
@@ -0,0 +1,746 @@
+/*
+ * 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.
+ */
+
+package org.apache.spark.ml.stat
+
+import breeze.{linalg => la}
+import breeze.linalg.{Vector => BV}
+import breeze.numerics
+
+import org.apache.spark.SparkException
+import org.apache.spark.annotation.Since
+import org.apache.spark.internal.Logging
+import org.apache.spark.ml.linalg.{DenseVector, SparseVector, Vector,
Vectors, VectorUDT}
+import org.apache.spark.sql.Column
+import org.apache.spark.sql.catalyst.InternalRow
+import org.apache.spark.sql.catalyst.expressions.{Expression,
UnsafeArrayData, UnsafeProjection, UnsafeRow}
+import
org.apache.spark.sql.catalyst.expressions.aggregate.{AggregateExpression,
Complete, TypedImperativeAggregate}
+import org.apache.spark.sql.types._
+
+
+/**
+ * A builder object that provides summary statistics about a given column.
+ *
+ * Users should not directly create such builders, but instead use one of
the methods in
+ * [[Summarizer]].
+ */
+@Since("2.2.0")
+abstract class SummaryBuilder {
+ /**
+ * Returns an aggregate object that contains the summary of the column
with the requested metrics.
+ * @param column a column that contains Vector object.
+ * @return an aggregate column that contains the statistics. The exact
content of this
+ * structure is determined during the creation of the builder.
+ */
+ @Since("2.2.0")
+ def summary(column: Column): Column
+}
+
+/**
+ * Tools for vectorized statistics on MLlib Vectors.
+ *
+ * The methods in this package provide various statistics for Vectors
contained inside DataFrames.
+ *
+ * This class lets users pick the statistics they would like to extract
for a given column. Here is
+ * an example in Scala:
+ * {{{
+ * val dataframe = ... // Some dataframe containing a feature column
+ * val allStats = dataframe.select(Summarizer.metrics("min",
"max").summary($"features"))
+ * val Row(min_, max_) = allStats.first()
+ * }}}
+ *
+ * If one wants to get a single metric, shortcuts are also available:
+ * {{{
+ * val meanDF = dataframe.select(Summarizer.mean($"features"))
+ * val Row(mean_) = meanDF.first()
+ * }}}
+ */
+@Since("2.2.0")
+object Summarizer extends Logging {
+
+ import SummaryBuilderImpl._
+
+ /**
+ * Given a list of metrics, provides a builder that it turns computes
metrics from a column.
+ *
+ * See the documentation of [[Summarizer]] for an example.
+ *
+ * The following metrics are accepted (case sensitive):
+ * - mean: a vector that contains the coefficient-wise mean.
+ * - variance: a vector tha contains the coefficient-wise variance.
+ * - count: the count of all vectors seen.
+ * - numNonzeros: a vector with the number of non-zeros for each
coefficients
+ * - max: the maximum for each coefficient.
+ * - min: the minimum for each coefficient.
+ * - normL2: the Euclidian norm for each coefficient.
+ * - normL1: the L1 norm of each coefficient (sum of the absolute
values).
+ * @param firstMetric the metric being provided
+ * @param metrics additional metrics that can be provided.
+ * @return a builder.
+ * @throws IllegalArgumentException if one of the metric names is not
understood.
+ */
+ @Since("2.2.0")
+ def metrics(firstMetric: String, metrics: String*): SummaryBuilder = {
+val (typedMetrics, computeMetrics) =
getRelevantMetrics(Seq(firstMetric) ++ metrics)
+new SummaryBuilderImpl(typedMetrics, computeMetrics)
+ }
+
+ def mean(col: Column): Column = getSingleMetric(col, "mean")
+
+ def variance(col: Column): Column = getSingleMetric(col, "variance")
[GitHub] spark pull request #17419: [SPARK-19634][ML] Multivariate summarizer - dataf...
Github user WeichenXu123 commented on a diff in the pull request:
https://github.com/apache/spark/pull/17419#discussion_r129173570
--- Diff: mllib/src/main/scala/org/apache/spark/ml/stat/Summarizer.scala ---
@@ -0,0 +1,799 @@
+/*
+ * 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.
+ */
+
+package org.apache.spark.ml.stat
+
+import breeze.{linalg => la}
+import breeze.linalg.{Vector => BV}
+import breeze.numerics
+
+import org.apache.spark.SparkException
+import org.apache.spark.annotation.Since
+import org.apache.spark.internal.Logging
+import org.apache.spark.ml.linalg.{DenseVector, SparseVector, Vector,
Vectors, VectorUDT}
+import org.apache.spark.sql.Column
+import org.apache.spark.sql.catalyst.InternalRow
+import org.apache.spark.sql.catalyst.expressions.{Expression,
UnsafeArrayData, UnsafeProjection, UnsafeRow}
+import
org.apache.spark.sql.catalyst.expressions.aggregate.{AggregateExpression,
Complete, TypedImperativeAggregate}
+import org.apache.spark.sql.types._
+
+
+/**
+ * A builder object that provides summary statistics about a given column.
+ *
+ * Users should not directly create such builders, but instead use one of
the methods in
+ * [[Summarizer]].
+ */
+@Since("2.2.0")
+abstract class SummaryBuilder {
+ /**
+ * Returns an aggregate object that contains the summary of the column
with the requested metrics.
+ * @param column a column that contains Vector object.
+ * @return an aggregate column that contains the statistics. The exact
content of this
+ * structure is determined during the creation of the builder.
+ */
+ @Since("2.2.0")
+ def summary(column: Column): Column
+}
+
+/**
+ * Tools for vectorized statistics on MLlib Vectors.
+ *
+ * The methods in this package provide various statistics for Vectors
contained inside DataFrames.
+ *
+ * This class lets users pick the statistics they would like to extract
for a given column. Here is
+ * an example in Scala:
+ * {{{
+ * val dataframe = ... // Some dataframe containing a feature column
+ * val allStats = dataframe.select(Summarizer.metrics("min",
"max").summary($"features"))
+ * val Row(min_, max_) = allStats.first()
+ * }}}
+ *
+ * If one wants to get a single metric, shortcuts are also available:
+ * {{{
+ * val meanDF = dataframe.select(Summarizer.mean($"features"))
+ * val Row(mean_) = meanDF.first()
+ * }}}
+ */
+@Since("2.2.0")
+object Summarizer extends Logging {
+
+ import SummaryBuilderImpl._
+
+ /**
+ * Given a list of metrics, provides a builder that it turns computes
metrics from a column.
+ *
+ * See the documentation of [[Summarizer]] for an example.
+ *
+ * The following metrics are accepted (case sensitive):
+ * - mean: a vector that contains the coefficient-wise mean.
+ * - variance: a vector tha contains the coefficient-wise variance.
+ * - count: the count of all vectors seen.
+ * - numNonzeros: a vector with the number of non-zeros for each
coefficients
+ * - max: the maximum for each coefficient.
+ * - min: the minimum for each coefficient.
+ * - normL2: the Euclidian norm for each coefficient.
+ * - normL1: the L1 norm of each coefficient (sum of the absolute
values).
+ * @param firstMetric the metric being provided
+ * @param metrics additional metrics that can be provided.
+ * @return a builder.
+ * @throws IllegalArgumentException if one of the metric names is not
understood.
+ */
+ @Since("2.2.0")
+ def metrics(firstMetric: String, metrics: String*): SummaryBuilder = {
+val (typedMetrics, computeMetrics) =
getRelevantMetrics(Seq(firstMetric) ++ metrics)
+new SummaryBuilderImpl(typedMetrics, computeMetrics)
+ }
+
+ def mean(col: Column): Column = getSingleMetric(col, "mean")
+
+ def variance(col: Column): Column = getSingleMetric(col,
[GitHub] spark pull request #17419: [SPARK-19634][ML] Multivariate summarizer - dataf...
Github user WeichenXu123 commented on a diff in the pull request:
https://github.com/apache/spark/pull/17419#discussion_r128429254
--- Diff: mllib/src/main/scala/org/apache/spark/ml/stat/Summarizer.scala ---
@@ -0,0 +1,799 @@
+/*
+ * 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.
+ */
+
+package org.apache.spark.ml.stat
+
+import breeze.{linalg => la}
+import breeze.linalg.{Vector => BV}
+import breeze.numerics
+
+import org.apache.spark.SparkException
+import org.apache.spark.annotation.Since
+import org.apache.spark.internal.Logging
+import org.apache.spark.ml.linalg.{DenseVector, SparseVector, Vector,
Vectors, VectorUDT}
+import org.apache.spark.sql.Column
+import org.apache.spark.sql.catalyst.InternalRow
+import org.apache.spark.sql.catalyst.expressions.{Expression,
UnsafeArrayData, UnsafeProjection, UnsafeRow}
+import
org.apache.spark.sql.catalyst.expressions.aggregate.{AggregateExpression,
Complete, TypedImperativeAggregate}
+import org.apache.spark.sql.types._
+
+
+/**
+ * A builder object that provides summary statistics about a given column.
+ *
+ * Users should not directly create such builders, but instead use one of
the methods in
+ * [[Summarizer]].
+ */
+@Since("2.2.0")
+abstract class SummaryBuilder {
+ /**
+ * Returns an aggregate object that contains the summary of the column
with the requested metrics.
+ * @param column a column that contains Vector object.
+ * @return an aggregate column that contains the statistics. The exact
content of this
+ * structure is determined during the creation of the builder.
+ */
+ @Since("2.2.0")
+ def summary(column: Column): Column
+}
+
+/**
+ * Tools for vectorized statistics on MLlib Vectors.
+ *
+ * The methods in this package provide various statistics for Vectors
contained inside DataFrames.
+ *
+ * This class lets users pick the statistics they would like to extract
for a given column. Here is
+ * an example in Scala:
+ * {{{
+ * val dataframe = ... // Some dataframe containing a feature column
+ * val allStats = dataframe.select(Summarizer.metrics("min",
"max").summary($"features"))
+ * val Row(min_, max_) = allStats.first()
+ * }}}
+ *
+ * If one wants to get a single metric, shortcuts are also available:
+ * {{{
+ * val meanDF = dataframe.select(Summarizer.mean($"features"))
+ * val Row(mean_) = meanDF.first()
+ * }}}
+ */
+@Since("2.2.0")
+object Summarizer extends Logging {
+
+ import SummaryBuilderImpl._
+
+ /**
+ * Given a list of metrics, provides a builder that it turns computes
metrics from a column.
+ *
+ * See the documentation of [[Summarizer]] for an example.
+ *
+ * The following metrics are accepted (case sensitive):
+ * - mean: a vector that contains the coefficient-wise mean.
+ * - variance: a vector tha contains the coefficient-wise variance.
+ * - count: the count of all vectors seen.
+ * - numNonzeros: a vector with the number of non-zeros for each
coefficients
+ * - max: the maximum for each coefficient.
+ * - min: the minimum for each coefficient.
+ * - normL2: the Euclidian norm for each coefficient.
+ * - normL1: the L1 norm of each coefficient (sum of the absolute
values).
+ * @param firstMetric the metric being provided
+ * @param metrics additional metrics that can be provided.
+ * @return a builder.
+ * @throws IllegalArgumentException if one of the metric names is not
understood.
+ */
+ @Since("2.2.0")
+ def metrics(firstMetric: String, metrics: String*): SummaryBuilder = {
+val (typedMetrics, computeMetrics) =
getRelevantMetrics(Seq(firstMetric) ++ metrics)
+new SummaryBuilderImpl(typedMetrics, computeMetrics)
+ }
+
+ def mean(col: Column): Column = getSingleMetric(col, "mean")
+
+ def variance(col: Column): Column = getSingleMetric(col,
[GitHub] spark pull request #17419: [SPARK-19634][ML] Multivariate summarizer - dataf...
Github user WeichenXu123 commented on a diff in the pull request:
https://github.com/apache/spark/pull/17419#discussion_r128428604
--- Diff: mllib/src/main/scala/org/apache/spark/ml/stat/Summarizer.scala ---
@@ -0,0 +1,746 @@
+/*
+ * 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.
+ */
+
+package org.apache.spark.ml.stat
+
+import breeze.{linalg => la}
+import breeze.linalg.{Vector => BV}
+import breeze.numerics
+
+import org.apache.spark.SparkException
+import org.apache.spark.annotation.Since
+import org.apache.spark.internal.Logging
+import org.apache.spark.ml.linalg.{DenseVector, SparseVector, Vector,
Vectors, VectorUDT}
+import org.apache.spark.sql.Column
+import org.apache.spark.sql.catalyst.InternalRow
+import org.apache.spark.sql.catalyst.expressions.{Expression,
UnsafeArrayData, UnsafeProjection, UnsafeRow}
+import
org.apache.spark.sql.catalyst.expressions.aggregate.{AggregateExpression,
Complete, TypedImperativeAggregate}
+import org.apache.spark.sql.types._
+
+
+/**
+ * A builder object that provides summary statistics about a given column.
+ *
+ * Users should not directly create such builders, but instead use one of
the methods in
+ * [[Summarizer]].
+ */
+@Since("2.2.0")
+abstract class SummaryBuilder {
+ /**
+ * Returns an aggregate object that contains the summary of the column
with the requested metrics.
+ * @param column a column that contains Vector object.
+ * @return an aggregate column that contains the statistics. The exact
content of this
+ * structure is determined during the creation of the builder.
+ */
+ @Since("2.2.0")
+ def summary(column: Column): Column
+}
+
+/**
+ * Tools for vectorized statistics on MLlib Vectors.
+ *
+ * The methods in this package provide various statistics for Vectors
contained inside DataFrames.
+ *
+ * This class lets users pick the statistics they would like to extract
for a given column. Here is
+ * an example in Scala:
+ * {{{
+ * val dataframe = ... // Some dataframe containing a feature column
+ * val allStats = dataframe.select(Summarizer.metrics("min",
"max").summary($"features"))
+ * val Row(min_, max_) = allStats.first()
+ * }}}
+ *
+ * If one wants to get a single metric, shortcuts are also available:
+ * {{{
+ * val meanDF = dataframe.select(Summarizer.mean($"features"))
+ * val Row(mean_) = meanDF.first()
+ * }}}
+ */
+@Since("2.2.0")
+object Summarizer extends Logging {
+
+ import SummaryBuilderImpl._
+
+ /**
+ * Given a list of metrics, provides a builder that it turns computes
metrics from a column.
+ *
+ * See the documentation of [[Summarizer]] for an example.
+ *
+ * The following metrics are accepted (case sensitive):
+ * - mean: a vector that contains the coefficient-wise mean.
+ * - variance: a vector tha contains the coefficient-wise variance.
+ * - count: the count of all vectors seen.
+ * - numNonzeros: a vector with the number of non-zeros for each
coefficients
+ * - max: the maximum for each coefficient.
+ * - min: the minimum for each coefficient.
+ * - normL2: the Euclidian norm for each coefficient.
+ * - normL1: the L1 norm of each coefficient (sum of the absolute
values).
+ * @param firstMetric the metric being provided
+ * @param metrics additional metrics that can be provided.
+ * @return a builder.
+ * @throws IllegalArgumentException if one of the metric names is not
understood.
+ */
+ @Since("2.2.0")
+ def metrics(firstMetric: String, metrics: String*): SummaryBuilder = {
+val (typedMetrics, computeMetrics) =
getRelevantMetrics(Seq(firstMetric) ++ metrics)
+new SummaryBuilderImpl(typedMetrics, computeMetrics)
+ }
+
+ def mean(col: Column): Column = getSingleMetric(col, "mean")
+
+ def variance(col: Column): Column = getSingleMetric(col,
[GitHub] spark pull request #17419: [SPARK-19634][ML] Multivariate summarizer - dataf...
Github user WeichenXu123 commented on a diff in the pull request:
https://github.com/apache/spark/pull/17419#discussion_r128429389
--- Diff:
mllib/src/test/scala/org/apache/spark/ml/stat/SummarizerSuite.scala ---
@@ -0,0 +1,406 @@
+/*
+ * 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.
+ */
+
+package org.apache.spark.ml.stat
+
+import org.scalatest.exceptions.TestFailedException
+
+import org.apache.spark.{SparkException, SparkFunSuite}
+import org.apache.spark.ml.linalg.{Vector, Vectors}
+import org.apache.spark.ml.stat.SummaryBuilderImpl.Buffer
+import org.apache.spark.ml.util.TestingUtils._
+import org.apache.spark.mllib.linalg.{Vector => OldVector, Vectors =>
OldVectors}
+import org.apache.spark.mllib.stat.{MultivariateOnlineSummarizer,
Statistics}
+import org.apache.spark.mllib.util.MLlibTestSparkContext
+import org.apache.spark.sql.{DataFrame, Row}
+import org.apache.spark.sql.catalyst.expressions.GenericRowWithSchema
+
+class SummarizerSuite extends SparkFunSuite with MLlibTestSparkContext {
+
+ import testImplicits._
+ import Summarizer._
+
+ private case class ExpectedMetrics(
+ mean: Seq[Double],
+ variance: Seq[Double],
+ count: Long,
+ numNonZeros: Seq[Long],
+ max: Seq[Double],
+ min: Seq[Double],
+ normL2: Seq[Double],
+ normL1: Seq[Double])
+
+ // The input is expected to be either a sparse vector, a dense vector or
an array of doubles
+ // (which will be converted to a dense vector)
+ // The expected is the list of all the known metrics.
+ //
+ // The tests take an list of input vectors and a list of all the summary
values that
+ // are expected for this input. They currently test against some fixed
subset of the
+ // metrics, but should be made fuzzy in the future.
+
+ private def testExample(name: String, input: Seq[Any], exp:
ExpectedMetrics): Unit = {
+def inputVec: Seq[Vector] = input.map {
+ case x: Array[Double @unchecked] => Vectors.dense(x)
+ case x: Seq[Double @unchecked] => Vectors.dense(x.toArray)
+ case x: Vector => x
+ case x => throw new Exception(x.toString)
+}
+
+val s = {
+ val s2 = new MultivariateOnlineSummarizer
+ inputVec.foreach(v => s2.add(OldVectors.fromML(v)))
+ s2
+}
+
+// Because the Spark context is reset between tests, we cannot hold a
reference onto it.
+def wrapped() = {
+ val df = sc.parallelize(inputVec).map(Tuple1.apply).toDF("features")
+ val c = df.col("features")
+ (df, c)
+}
+
+registerTest(s"$name - mean only") {
+ val (df, c) = wrapped()
+ compare(df.select(metrics("mean").summary(c), mean(c)),
Seq(Row(exp.mean), s.mean))
+}
+
+registerTest(s"$name - mean only (direct)") {
+ val (df, c) = wrapped()
+ compare(df.select(mean(c)), Seq(exp.mean))
+}
+
+registerTest(s"$name - variance only") {
+ val (df, c) = wrapped()
+ compare(df.select(metrics("variance").summary(c), variance(c)),
+Seq(Row(exp.variance), s.variance))
+}
+
+registerTest(s"$name - variance only (direct)") {
+ val (df, c) = wrapped()
+ compare(df.select(variance(c)), Seq(s.variance))
+}
+
+registerTest(s"$name - count only") {
+ val (df, c) = wrapped()
+ compare(df.select(metrics("count").summary(c), count(c)),
+Seq(Row(exp.count), exp.count))
+}
+
+registerTest(s"$name - count only (direct)") {
+ val (df, c) = wrapped()
+ compare(df.select(count(c)),
+Seq(exp.count))
+}
+
+registerTest(s"$name - numNonZeros only") {
+ val (df, c) = wrapped()
+ compare(df.select(metrics("numNonZeros").summary(c), numNonZeros(c)),
+Seq(Row(exp.numNonZeros), exp.numNonZeros))
+}
+
+registerTest(s"$name
[GitHub] spark pull request #17419: [SPARK-19634][ML] Multivariate summarizer - dataf...
Github user WeichenXu123 commented on a diff in the pull request:
https://github.com/apache/spark/pull/17419#discussion_r128428434
--- Diff: mllib/src/main/scala/org/apache/spark/ml/stat/Summarizer.scala ---
@@ -0,0 +1,746 @@
+/*
+ * 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.
+ */
+
+package org.apache.spark.ml.stat
+
+import breeze.{linalg => la}
+import breeze.linalg.{Vector => BV}
+import breeze.numerics
+
+import org.apache.spark.SparkException
+import org.apache.spark.annotation.Since
+import org.apache.spark.internal.Logging
+import org.apache.spark.ml.linalg.{DenseVector, SparseVector, Vector,
Vectors, VectorUDT}
+import org.apache.spark.sql.Column
+import org.apache.spark.sql.catalyst.InternalRow
+import org.apache.spark.sql.catalyst.expressions.{Expression,
UnsafeArrayData, UnsafeProjection, UnsafeRow}
+import
org.apache.spark.sql.catalyst.expressions.aggregate.{AggregateExpression,
Complete, TypedImperativeAggregate}
+import org.apache.spark.sql.types._
+
+
+/**
+ * A builder object that provides summary statistics about a given column.
+ *
+ * Users should not directly create such builders, but instead use one of
the methods in
+ * [[Summarizer]].
+ */
+@Since("2.2.0")
+abstract class SummaryBuilder {
+ /**
+ * Returns an aggregate object that contains the summary of the column
with the requested metrics.
+ * @param column a column that contains Vector object.
+ * @return an aggregate column that contains the statistics. The exact
content of this
+ * structure is determined during the creation of the builder.
+ */
+ @Since("2.2.0")
+ def summary(column: Column): Column
+}
+
+/**
+ * Tools for vectorized statistics on MLlib Vectors.
+ *
+ * The methods in this package provide various statistics for Vectors
contained inside DataFrames.
+ *
+ * This class lets users pick the statistics they would like to extract
for a given column. Here is
+ * an example in Scala:
+ * {{{
+ * val dataframe = ... // Some dataframe containing a feature column
+ * val allStats = dataframe.select(Summarizer.metrics("min",
"max").summary($"features"))
+ * val Row(min_, max_) = allStats.first()
+ * }}}
+ *
+ * If one wants to get a single metric, shortcuts are also available:
+ * {{{
+ * val meanDF = dataframe.select(Summarizer.mean($"features"))
+ * val Row(mean_) = meanDF.first()
+ * }}}
+ */
+@Since("2.2.0")
+object Summarizer extends Logging {
+
+ import SummaryBuilderImpl._
+
+ /**
+ * Given a list of metrics, provides a builder that it turns computes
metrics from a column.
+ *
+ * See the documentation of [[Summarizer]] for an example.
+ *
+ * The following metrics are accepted (case sensitive):
+ * - mean: a vector that contains the coefficient-wise mean.
+ * - variance: a vector tha contains the coefficient-wise variance.
+ * - count: the count of all vectors seen.
+ * - numNonzeros: a vector with the number of non-zeros for each
coefficients
+ * - max: the maximum for each coefficient.
+ * - min: the minimum for each coefficient.
+ * - normL2: the Euclidian norm for each coefficient.
+ * - normL1: the L1 norm of each coefficient (sum of the absolute
values).
+ * @param firstMetric the metric being provided
+ * @param metrics additional metrics that can be provided.
+ * @return a builder.
+ * @throws IllegalArgumentException if one of the metric names is not
understood.
+ */
+ @Since("2.2.0")
+ def metrics(firstMetric: String, metrics: String*): SummaryBuilder = {
+val (typedMetrics, computeMetrics) =
getRelevantMetrics(Seq(firstMetric) ++ metrics)
+new SummaryBuilderImpl(typedMetrics, computeMetrics)
+ }
+
+ def mean(col: Column): Column = getSingleMetric(col, "mean")
+
+ def variance(col: Column): Column = getSingleMetric(col,
[GitHub] spark pull request #17419: [SPARK-19634][ML] Multivariate summarizer - dataf...
Github user cloud-fan commented on a diff in the pull request:
https://github.com/apache/spark/pull/17419#discussion_r115306420
--- Diff: mllib/src/main/scala/org/apache/spark/ml/stat/Summarizer.scala ---
@@ -0,0 +1,799 @@
+/*
+ * 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.
+ */
+
+package org.apache.spark.ml.stat
+
+import breeze.{linalg => la}
+import breeze.linalg.{Vector => BV}
+import breeze.numerics
+
+import org.apache.spark.SparkException
+import org.apache.spark.annotation.Since
+import org.apache.spark.internal.Logging
+import org.apache.spark.ml.linalg.{DenseVector, SparseVector, Vector,
Vectors, VectorUDT}
+import org.apache.spark.sql.Column
+import org.apache.spark.sql.catalyst.InternalRow
+import org.apache.spark.sql.catalyst.expressions.{Expression,
UnsafeArrayData, UnsafeProjection, UnsafeRow}
+import
org.apache.spark.sql.catalyst.expressions.aggregate.{AggregateExpression,
Complete, TypedImperativeAggregate}
+import org.apache.spark.sql.types._
+
+
+/**
+ * A builder object that provides summary statistics about a given column.
+ *
+ * Users should not directly create such builders, but instead use one of
the methods in
+ * [[Summarizer]].
+ */
+@Since("2.2.0")
+abstract class SummaryBuilder {
+ /**
+ * Returns an aggregate object that contains the summary of the column
with the requested metrics.
+ * @param column a column that contains Vector object.
+ * @return an aggregate column that contains the statistics. The exact
content of this
+ * structure is determined during the creation of the builder.
+ */
+ @Since("2.2.0")
+ def summary(column: Column): Column
+}
+
+/**
+ * Tools for vectorized statistics on MLlib Vectors.
+ *
+ * The methods in this package provide various statistics for Vectors
contained inside DataFrames.
+ *
+ * This class lets users pick the statistics they would like to extract
for a given column. Here is
+ * an example in Scala:
+ * {{{
+ * val dataframe = ... // Some dataframe containing a feature column
+ * val allStats = dataframe.select(Summarizer.metrics("min",
"max").summary($"features"))
+ * val Row(min_, max_) = allStats.first()
+ * }}}
+ *
+ * If one wants to get a single metric, shortcuts are also available:
+ * {{{
+ * val meanDF = dataframe.select(Summarizer.mean($"features"))
+ * val Row(mean_) = meanDF.first()
+ * }}}
+ */
+@Since("2.2.0")
+object Summarizer extends Logging {
+
+ import SummaryBuilderImpl._
+
+ /**
+ * Given a list of metrics, provides a builder that it turns computes
metrics from a column.
+ *
+ * See the documentation of [[Summarizer]] for an example.
+ *
+ * The following metrics are accepted (case sensitive):
+ * - mean: a vector that contains the coefficient-wise mean.
+ * - variance: a vector tha contains the coefficient-wise variance.
+ * - count: the count of all vectors seen.
+ * - numNonzeros: a vector with the number of non-zeros for each
coefficients
+ * - max: the maximum for each coefficient.
+ * - min: the minimum for each coefficient.
+ * - normL2: the Euclidian norm for each coefficient.
+ * - normL1: the L1 norm of each coefficient (sum of the absolute
values).
+ * @param firstMetric the metric being provided
+ * @param metrics additional metrics that can be provided.
+ * @return a builder.
+ * @throws IllegalArgumentException if one of the metric names is not
understood.
+ */
+ @Since("2.2.0")
+ def metrics(firstMetric: String, metrics: String*): SummaryBuilder = {
+val (typedMetrics, computeMetrics) =
getRelevantMetrics(Seq(firstMetric) ++ metrics)
+new SummaryBuilderImpl(typedMetrics, computeMetrics)
+ }
+
+ def mean(col: Column): Column = getSingleMetric(col, "mean")
+
+ def variance(col: Column): Column = getSingleMetric(col, "variance")
[GitHub] spark pull request #17419: [SPARK-19634][ML] Multivariate summarizer - dataf...
Github user thunterdb commented on a diff in the pull request:
https://github.com/apache/spark/pull/17419#discussion_r109063248
--- Diff: mllib/src/main/scala/org/apache/spark/ml/stat/Summarizer.scala ---
@@ -0,0 +1,746 @@
+/*
+ * 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.
+ */
+
+package org.apache.spark.ml.stat
+
+import breeze.{linalg => la}
+import breeze.linalg.{Vector => BV}
+import breeze.numerics
+
+import org.apache.spark.SparkException
+import org.apache.spark.annotation.Since
+import org.apache.spark.internal.Logging
+import org.apache.spark.ml.linalg.{DenseVector, SparseVector, Vector,
Vectors, VectorUDT}
+import org.apache.spark.sql.Column
+import org.apache.spark.sql.catalyst.InternalRow
+import org.apache.spark.sql.catalyst.expressions.{Expression,
UnsafeArrayData, UnsafeProjection, UnsafeRow}
+import
org.apache.spark.sql.catalyst.expressions.aggregate.{AggregateExpression,
Complete, TypedImperativeAggregate}
+import org.apache.spark.sql.types._
+
+
+/**
+ * A builder object that provides summary statistics about a given column.
+ *
+ * Users should not directly create such builders, but instead use one of
the methods in
+ * [[Summarizer]].
+ */
+@Since("2.2.0")
+abstract class SummaryBuilder {
+ /**
+ * Returns an aggregate object that contains the summary of the column
with the requested metrics.
+ * @param column a column that contains Vector object.
+ * @return an aggregate column that contains the statistics. The exact
content of this
+ * structure is determined during the creation of the builder.
+ */
+ @Since("2.2.0")
+ def summary(column: Column): Column
+}
+
+/**
+ * Tools for vectorized statistics on MLlib Vectors.
+ *
+ * The methods in this package provide various statistics for Vectors
contained inside DataFrames.
+ *
+ * This class lets users pick the statistics they would like to extract
for a given column. Here is
+ * an example in Scala:
+ * {{{
+ * val dataframe = ... // Some dataframe containing a feature column
+ * val allStats = dataframe.select(Summarizer.metrics("min",
"max").summary($"features"))
+ * val Row(min_, max_) = allStats.first()
+ * }}}
+ *
+ * If one wants to get a single metric, shortcuts are also available:
+ * {{{
+ * val meanDF = dataframe.select(Summarizer.mean($"features"))
+ * val Row(mean_) = meanDF.first()
+ * }}}
+ */
+@Since("2.2.0")
+object Summarizer extends Logging {
+
+ import SummaryBuilderImpl._
+
+ /**
+ * Given a list of metrics, provides a builder that it turns computes
metrics from a column.
+ *
+ * See the documentation of [[Summarizer]] for an example.
+ *
+ * The following metrics are accepted (case sensitive):
+ * - mean: a vector that contains the coefficient-wise mean.
+ * - variance: a vector tha contains the coefficient-wise variance.
+ * - count: the count of all vectors seen.
+ * - numNonzeros: a vector with the number of non-zeros for each
coefficients
+ * - max: the maximum for each coefficient.
+ * - min: the minimum for each coefficient.
+ * - normL2: the Euclidian norm for each coefficient.
+ * - normL1: the L1 norm of each coefficient (sum of the absolute
values).
+ * @param firstMetric the metric being provided
+ * @param metrics additional metrics that can be provided.
+ * @return a builder.
+ * @throws IllegalArgumentException if one of the metric names is not
understood.
+ */
+ @Since("2.2.0")
+ def metrics(firstMetric: String, metrics: String*): SummaryBuilder = {
+val (typedMetrics, computeMetrics) =
getRelevantMetrics(Seq(firstMetric) ++ metrics)
+new SummaryBuilderImpl(typedMetrics, computeMetrics)
+ }
+
+ def mean(col: Column): Column = getSingleMetric(col, "mean")
+
+ def variance(col: Column): Column = getSingleMetric(col, "variance")
[GitHub] spark pull request #17419: [SPARK-19634][ML] Multivariate summarizer - dataf...
Github user viirya commented on a diff in the pull request:
https://github.com/apache/spark/pull/17419#discussion_r108858117
--- Diff: mllib/src/main/scala/org/apache/spark/ml/stat/Summarizer.scala ---
@@ -0,0 +1,746 @@
+/*
+ * 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.
+ */
+
+package org.apache.spark.ml.stat
+
+import breeze.{linalg => la}
+import breeze.linalg.{Vector => BV}
+import breeze.numerics
+
+import org.apache.spark.SparkException
+import org.apache.spark.annotation.Since
+import org.apache.spark.internal.Logging
+import org.apache.spark.ml.linalg.{DenseVector, SparseVector, Vector,
Vectors, VectorUDT}
+import org.apache.spark.sql.Column
+import org.apache.spark.sql.catalyst.InternalRow
+import org.apache.spark.sql.catalyst.expressions.{Expression,
UnsafeArrayData, UnsafeProjection, UnsafeRow}
+import
org.apache.spark.sql.catalyst.expressions.aggregate.{AggregateExpression,
Complete, TypedImperativeAggregate}
+import org.apache.spark.sql.types._
+
+
+/**
+ * A builder object that provides summary statistics about a given column.
+ *
+ * Users should not directly create such builders, but instead use one of
the methods in
+ * [[Summarizer]].
+ */
+@Since("2.2.0")
+abstract class SummaryBuilder {
+ /**
+ * Returns an aggregate object that contains the summary of the column
with the requested metrics.
+ * @param column a column that contains Vector object.
+ * @return an aggregate column that contains the statistics. The exact
content of this
+ * structure is determined during the creation of the builder.
+ */
+ @Since("2.2.0")
+ def summary(column: Column): Column
+}
+
+/**
+ * Tools for vectorized statistics on MLlib Vectors.
+ *
+ * The methods in this package provide various statistics for Vectors
contained inside DataFrames.
+ *
+ * This class lets users pick the statistics they would like to extract
for a given column. Here is
+ * an example in Scala:
+ * {{{
+ * val dataframe = ... // Some dataframe containing a feature column
+ * val allStats = dataframe.select(Summarizer.metrics("min",
"max").summary($"features"))
+ * val Row(min_, max_) = allStats.first()
+ * }}}
+ *
+ * If one wants to get a single metric, shortcuts are also available:
+ * {{{
+ * val meanDF = dataframe.select(Summarizer.mean($"features"))
+ * val Row(mean_) = meanDF.first()
+ * }}}
+ */
+@Since("2.2.0")
+object Summarizer extends Logging {
+
+ import SummaryBuilderImpl._
+
+ /**
+ * Given a list of metrics, provides a builder that it turns computes
metrics from a column.
+ *
+ * See the documentation of [[Summarizer]] for an example.
+ *
+ * The following metrics are accepted (case sensitive):
+ * - mean: a vector that contains the coefficient-wise mean.
+ * - variance: a vector tha contains the coefficient-wise variance.
+ * - count: the count of all vectors seen.
+ * - numNonzeros: a vector with the number of non-zeros for each
coefficients
+ * - max: the maximum for each coefficient.
+ * - min: the minimum for each coefficient.
+ * - normL2: the Euclidian norm for each coefficient.
+ * - normL1: the L1 norm of each coefficient (sum of the absolute
values).
+ * @param firstMetric the metric being provided
+ * @param metrics additional metrics that can be provided.
+ * @return a builder.
+ * @throws IllegalArgumentException if one of the metric names is not
understood.
+ */
+ @Since("2.2.0")
+ def metrics(firstMetric: String, metrics: String*): SummaryBuilder = {
+val (typedMetrics, computeMetrics) =
getRelevantMetrics(Seq(firstMetric) ++ metrics)
+new SummaryBuilderImpl(typedMetrics, computeMetrics)
+ }
+
+ def mean(col: Column): Column = getSingleMetric(col, "mean")
+
+ def variance(col: Column): Column = getSingleMetric(col, "variance")
[GitHub] spark pull request #17419: [SPARK-19634][ML] Multivariate summarizer - dataf...
Github user viirya commented on a diff in the pull request:
https://github.com/apache/spark/pull/17419#discussion_r108856798
--- Diff: mllib/src/main/scala/org/apache/spark/ml/stat/Summarizer.scala ---
@@ -0,0 +1,746 @@
+/*
+ * 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.
+ */
+
+package org.apache.spark.ml.stat
+
+import breeze.{linalg => la}
+import breeze.linalg.{Vector => BV}
+import breeze.numerics
+
+import org.apache.spark.SparkException
+import org.apache.spark.annotation.Since
+import org.apache.spark.internal.Logging
+import org.apache.spark.ml.linalg.{DenseVector, SparseVector, Vector,
Vectors, VectorUDT}
+import org.apache.spark.sql.Column
+import org.apache.spark.sql.catalyst.InternalRow
+import org.apache.spark.sql.catalyst.expressions.{Expression,
UnsafeArrayData, UnsafeProjection, UnsafeRow}
+import
org.apache.spark.sql.catalyst.expressions.aggregate.{AggregateExpression,
Complete, TypedImperativeAggregate}
+import org.apache.spark.sql.types._
+
+
+/**
+ * A builder object that provides summary statistics about a given column.
+ *
+ * Users should not directly create such builders, but instead use one of
the methods in
+ * [[Summarizer]].
+ */
+@Since("2.2.0")
+abstract class SummaryBuilder {
+ /**
+ * Returns an aggregate object that contains the summary of the column
with the requested metrics.
+ * @param column a column that contains Vector object.
+ * @return an aggregate column that contains the statistics. The exact
content of this
+ * structure is determined during the creation of the builder.
+ */
+ @Since("2.2.0")
+ def summary(column: Column): Column
+}
+
+/**
+ * Tools for vectorized statistics on MLlib Vectors.
+ *
+ * The methods in this package provide various statistics for Vectors
contained inside DataFrames.
+ *
+ * This class lets users pick the statistics they would like to extract
for a given column. Here is
+ * an example in Scala:
+ * {{{
+ * val dataframe = ... // Some dataframe containing a feature column
+ * val allStats = dataframe.select(Summarizer.metrics("min",
"max").summary($"features"))
+ * val Row(min_, max_) = allStats.first()
+ * }}}
+ *
+ * If one wants to get a single metric, shortcuts are also available:
+ * {{{
+ * val meanDF = dataframe.select(Summarizer.mean($"features"))
+ * val Row(mean_) = meanDF.first()
+ * }}}
+ */
+@Since("2.2.0")
+object Summarizer extends Logging {
+
+ import SummaryBuilderImpl._
+
+ /**
+ * Given a list of metrics, provides a builder that it turns computes
metrics from a column.
+ *
+ * See the documentation of [[Summarizer]] for an example.
+ *
+ * The following metrics are accepted (case sensitive):
+ * - mean: a vector that contains the coefficient-wise mean.
+ * - variance: a vector tha contains the coefficient-wise variance.
+ * - count: the count of all vectors seen.
+ * - numNonzeros: a vector with the number of non-zeros for each
coefficients
+ * - max: the maximum for each coefficient.
+ * - min: the minimum for each coefficient.
+ * - normL2: the Euclidian norm for each coefficient.
+ * - normL1: the L1 norm of each coefficient (sum of the absolute
values).
+ * @param firstMetric the metric being provided
+ * @param metrics additional metrics that can be provided.
+ * @return a builder.
+ * @throws IllegalArgumentException if one of the metric names is not
understood.
+ */
+ @Since("2.2.0")
+ def metrics(firstMetric: String, metrics: String*): SummaryBuilder = {
+val (typedMetrics, computeMetrics) =
getRelevantMetrics(Seq(firstMetric) ++ metrics)
+new SummaryBuilderImpl(typedMetrics, computeMetrics)
+ }
+
+ def mean(col: Column): Column = getSingleMetric(col, "mean")
+
+ def variance(col: Column): Column = getSingleMetric(col, "variance")
[GitHub] spark pull request #17419: [SPARK-19634][ML] Multivariate summarizer - dataf...
Github user viirya commented on a diff in the pull request:
https://github.com/apache/spark/pull/17419#discussion_r108840757
--- Diff:
mllib/src/test/scala/org/apache/spark/ml/stat/SummarizerSuite.scala ---
@@ -0,0 +1,399 @@
+/*
+ * 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.
+ */
+
+package org.apache.spark.ml.stat
+
+import org.scalatest.exceptions.TestFailedException
+
+import org.apache.spark.{SparkException, SparkFunSuite}
+import org.apache.spark.ml.linalg.{Vector, Vectors}
+import org.apache.spark.ml.stat.SummaryBuilderImpl.Buffer
+import org.apache.spark.ml.util.TestingUtils._
+import org.apache.spark.mllib.linalg.{Vector => OldVector, Vectors =>
OldVectors}
+import org.apache.spark.mllib.stat.{MultivariateOnlineSummarizer,
Statistics}
+import org.apache.spark.mllib.util.MLlibTestSparkContext
+import org.apache.spark.sql.{DataFrame, Row}
+import org.apache.spark.sql.catalyst.expressions.GenericRowWithSchema
+
+class SummarizerSuite extends SparkFunSuite with MLlibTestSparkContext {
+
+ import testImplicits._
+ import Summarizer._
+
+ private case class ExpectedMetrics(
+ mean: Seq[Double],
+ variance: Seq[Double],
+ count: Long,
+ numNonZeros: Seq[Long],
+ max: Seq[Double],
+ min: Seq[Double],
+ normL2: Seq[Double],
+ normL1: Seq[Double])
+
+ // The input is expected to be either a sparse vector, a dense vector or
an array of doubles
+ // (which will be converted to a dense vector)
+ // The expected is the list of all the known metrics.
+ //
+ // The tests take an list of input vectors and a list of all the summary
values that
+ // are expected for this input. They currently test against some fixed
subset of the
+ // metrics, but should be made fuzzy in the future.
+
+ private def testExample(name: String, input: Seq[Any], exp:
ExpectedMetrics): Unit = {
+def inputVec: Seq[Vector] = input.map {
+ case x: Array[Double @unchecked] => Vectors.dense(x)
+ case x: Seq[Double @unchecked] => Vectors.dense(x.toArray)
+ case x: Vector => x
+ case x => throw new Exception(x.toString)
+}
+
+val s = {
+ val s2 = new MultivariateOnlineSummarizer
+ inputVec.foreach(v => s2.add(OldVectors.fromML(v)))
+ s2
+}
+
+// Because the Spark context is reset between tests, we cannot hold a
reference onto it.
+def wrapped() = {
+ val df = sc.parallelize(inputVec).map(Tuple1.apply).toDF("features")
+ val c = df.col("features")
+ (df, c)
+}
+
+registerTest(s"$name - mean only") {
+ val (df, c) = wrapped()
+ compare(df.select(metrics("mean").summary(c), mean(c)),
Seq(Row(exp.mean), s.mean))
+}
+
+registerTest(s"$name - mean only (direct)") {
+ val (df, c) = wrapped()
+ compare(df.select(mean(c)), Seq(exp.mean))
+}
+
+registerTest(s"$name - variance only") {
+ val (df, c) = wrapped()
+ compare(df.select(metrics("variance").summary(c), variance(c)),
+Seq(Row(exp.variance), s.variance))
+}
+
+registerTest(s"$name - variance only (direct)") {
+ val (df, c) = wrapped()
+ compare(df.select(variance(c)), Seq(s.variance))
+}
+
+registerTest(s"$name - count only") {
+ val (df, c) = wrapped()
+ compare(df.select(metrics("count").summary(c), count(c)),
+Seq(Row(exp.count), exp.count))
+}
+
+registerTest(s"$name - count only (direct)") {
+ val (df, c) = wrapped()
+ compare(df.select(count(c)),
+Seq(exp.count))
+}
+
+registerTest(s"$name - numNonZeros only") {
+ val (df, c) = wrapped()
+ compare(df.select(metrics("numNonZeros").summary(c), numNonZeros(c)),
+Seq(Row(exp.numNonZeros), exp.numNonZeros))
+}
+
+registerTest(s"$name -
[GitHub] spark pull request #17419: [SPARK-19634][ML] Multivariate summarizer - dataf...
Github user viirya commented on a diff in the pull request:
https://github.com/apache/spark/pull/17419#discussion_r108840709
--- Diff:
mllib/src/test/scala/org/apache/spark/ml/stat/SummarizerSuite.scala ---
@@ -0,0 +1,338 @@
+/*
+ * 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.
+ */
+
+package org.apache.spark.ml.stat
+
+import org.scalatest.exceptions.TestFailedException
+
+import org.apache.spark.{SparkException, SparkFunSuite}
+import org.apache.spark.ml.linalg.{Vector, Vectors}
+import org.apache.spark.ml.stat.SummaryBuilderImpl.Buffer
+import org.apache.spark.ml.util.TestingUtils._
+import org.apache.spark.mllib.linalg.{Vector => OldVector, Vectors =>
OldVectors}
+import org.apache.spark.mllib.stat.MultivariateOnlineSummarizer
+import org.apache.spark.mllib.util.MLlibTestSparkContext
+import org.apache.spark.sql.{DataFrame, Row}
+import org.apache.spark.sql.catalyst.expressions.GenericRowWithSchema
+
+class SummarizerSuite extends SparkFunSuite with MLlibTestSparkContext {
+
+ import testImplicits._
+ import Summarizer._
+
+ private case class ExpectedMetrics(
+ mean: Seq[Double],
+ variance: Seq[Double],
+ count: Long,
+ numNonZeros: Seq[Long],
+ max: Seq[Double],
+ min: Seq[Double],
+ normL2: Seq[Double],
+ normL1: Seq[Double])
+
+ // The input is expected to be either a sparse vector, a dense vector or
an array of doubles
+ // (which will be converted to a dense vector)
+ // The expected is the list of all the known metrics.
+ //
+ // The tests take an list of input vectors and a list of all the summary
values that
+ // are expected for this input. They currently test against some fixed
subset of the
+ // metrics, but should be made fuzzy in the future.
+
+ private def testExample(name: String, input: Seq[Any], exp:
ExpectedMetrics): Unit = {
+def inputVec: Seq[Vector] = input.map {
+ case x: Array[Double @unchecked] => Vectors.dense(x)
+ case x: Seq[Double @unchecked] => Vectors.dense(x.toArray)
+ case x: Vector => x
+ case x => throw new Exception(x.toString)
+}
+
+val s = {
+ val s2 = new MultivariateOnlineSummarizer
+ inputVec.foreach(v => s2.add(OldVectors.fromML(v)))
+ s2
+}
+
+// Because the Spark context is reset between tests, we cannot hold a
reference onto it.
+def wrapped() = {
+ val df = sc.parallelize(inputVec).map(Tuple1.apply).toDF("features")
+ val c = df.col("features")
+ (df, c)
+}
+
+registerTest(s"$name - mean only") {
+ val (df, c) = wrapped()
+ compare(df.select(metrics("mean").summary(c), mean(c)),
Seq(Row(exp.mean), s.mean))
+}
+
+registerTest(s"$name - mean only (direct)") {
+ val (df, c) = wrapped()
+ compare(df.select(mean(c)), Seq(exp.mean))
+}
+
+registerTest(s"$name - variance only") {
+ val (df, c) = wrapped()
+ compare(df.select(metrics("variance").summary(c), variance(c)),
+Seq(Row(exp.variance), s.variance))
+}
+
+registerTest(s"$name - variance only (direct)") {
+ val (df, c) = wrapped()
+ compare(df.select(variance(c)), Seq(s.variance))
+}
+
+registerTest(s"$name - count only") {
+ val (df, c) = wrapped()
+ compare(df.select(metrics("count").summary(c), count(c)),
+Seq(Row(exp.count), exp.count))
+}
+
+registerTest(s"$name - count only (direct)") {
+ val (df, c) = wrapped()
+ compare(df.select(count(c)),
+Seq(exp.count))
+}
+
+registerTest(s"$name - numNonZeros only") {
+ val (df, c) = wrapped()
+ compare(df.select(metrics("numNonZeros").summary(c), numNonZeros(c)),
+Seq(Row(exp.numNonZeros), exp.numNonZeros))
+}
+
+registerTest(s"$name - numNonZeros only
[GitHub] spark pull request #17419: [SPARK-19634][ML] Multivariate summarizer - dataf...
Github user viirya commented on a diff in the pull request:
https://github.com/apache/spark/pull/17419#discussion_r108838518
--- Diff: mllib/src/main/scala/org/apache/spark/ml/stat/Summarizer.scala ---
@@ -0,0 +1,746 @@
+/*
+ * 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.
+ */
+
+package org.apache.spark.ml.stat
+
+import breeze.{linalg => la}
+import breeze.linalg.{Vector => BV}
+import breeze.numerics
+
+import org.apache.spark.SparkException
+import org.apache.spark.annotation.Since
+import org.apache.spark.internal.Logging
+import org.apache.spark.ml.linalg.{DenseVector, SparseVector, Vector,
Vectors, VectorUDT}
+import org.apache.spark.sql.Column
+import org.apache.spark.sql.catalyst.InternalRow
+import org.apache.spark.sql.catalyst.expressions.{Expression,
UnsafeArrayData, UnsafeProjection, UnsafeRow}
+import
org.apache.spark.sql.catalyst.expressions.aggregate.{AggregateExpression,
Complete, TypedImperativeAggregate}
+import org.apache.spark.sql.types._
+
+
+/**
+ * A builder object that provides summary statistics about a given column.
+ *
+ * Users should not directly create such builders, but instead use one of
the methods in
+ * [[Summarizer]].
+ */
+@Since("2.2.0")
+abstract class SummaryBuilder {
+ /**
+ * Returns an aggregate object that contains the summary of the column
with the requested metrics.
+ * @param column a column that contains Vector object.
+ * @return an aggregate column that contains the statistics. The exact
content of this
+ * structure is determined during the creation of the builder.
+ */
+ @Since("2.2.0")
+ def summary(column: Column): Column
+}
+
+/**
+ * Tools for vectorized statistics on MLlib Vectors.
+ *
+ * The methods in this package provide various statistics for Vectors
contained inside DataFrames.
+ *
+ * This class lets users pick the statistics they would like to extract
for a given column. Here is
+ * an example in Scala:
+ * {{{
+ * val dataframe = ... // Some dataframe containing a feature column
+ * val allStats = dataframe.select(Summarizer.metrics("min",
"max").summary($"features"))
+ * val Row(min_, max_) = allStats.first()
+ * }}}
+ *
+ * If one wants to get a single metric, shortcuts are also available:
+ * {{{
+ * val meanDF = dataframe.select(Summarizer.mean($"features"))
+ * val Row(mean_) = meanDF.first()
+ * }}}
+ */
+@Since("2.2.0")
+object Summarizer extends Logging {
+
+ import SummaryBuilderImpl._
+
+ /**
+ * Given a list of metrics, provides a builder that it turns computes
metrics from a column.
+ *
+ * See the documentation of [[Summarizer]] for an example.
+ *
+ * The following metrics are accepted (case sensitive):
+ * - mean: a vector that contains the coefficient-wise mean.
+ * - variance: a vector tha contains the coefficient-wise variance.
+ * - count: the count of all vectors seen.
+ * - numNonzeros: a vector with the number of non-zeros for each
coefficients
+ * - max: the maximum for each coefficient.
+ * - min: the minimum for each coefficient.
+ * - normL2: the Euclidian norm for each coefficient.
+ * - normL1: the L1 norm of each coefficient (sum of the absolute
values).
+ * @param firstMetric the metric being provided
+ * @param metrics additional metrics that can be provided.
+ * @return a builder.
+ * @throws IllegalArgumentException if one of the metric names is not
understood.
+ */
+ @Since("2.2.0")
+ def metrics(firstMetric: String, metrics: String*): SummaryBuilder = {
+val (typedMetrics, computeMetrics) =
getRelevantMetrics(Seq(firstMetric) ++ metrics)
+new SummaryBuilderImpl(typedMetrics, computeMetrics)
+ }
+
+ def mean(col: Column): Column = getSingleMetric(col, "mean")
+
+ def variance(col: Column): Column = getSingleMetric(col, "variance")
[GitHub] spark pull request #17419: [SPARK-19634][ML] Multivariate summarizer - dataf...
Github user kiszk commented on a diff in the pull request:
https://github.com/apache/spark/pull/17419#discussion_r108746608
--- Diff:
mllib/src/test/scala/org/apache/spark/ml/stat/SummarizerSuite.scala ---
@@ -335,4 +335,65 @@ class SummarizerSuite extends SparkFunSuite with
MLlibTestSparkContext {
assert(Buffer.totalCount(summarizer) === 6)
}
+ // TODO: this test should not be committed. It is here to isolate some
performance hotspots.
+ test("perf test") {
+val n = 1000
+val rdd1 = sc.parallelize(1 to n).map { idx =>
+ OldVectors.dense(idx.toDouble)
+}
+val trieouts = 10
--- End diff --
When results are stabilized, I think that it would be good to keep results
with ``ignore("benchmark")`` as [other benchmarks
does](https://github.com/apache/spark/blob/master/sql/core/src/test/scala/org/apache/spark/sql/execution/benchmark/MiscBenchmark.scala).
---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If your project does not have this feature
enabled and wishes so, or if the feature is enabled but not working, please
contact infrastructure at infrastruct...@apache.org or file a JIRA ticket
with INFRA.
---
-
To unsubscribe, e-mail: reviews-unsubscr...@spark.apache.org
For additional commands, e-mail: reviews-h...@spark.apache.org
[GitHub] spark pull request #17419: [SPARK-19634][ML] Multivariate summarizer - dataf...
Github user thunterdb commented on a diff in the pull request:
https://github.com/apache/spark/pull/17419#discussion_r108743634
--- Diff:
mllib/src/test/scala/org/apache/spark/ml/stat/SummarizerSuite.scala ---
@@ -335,4 +335,65 @@ class SummarizerSuite extends SparkFunSuite with
MLlibTestSparkContext {
assert(Buffer.totalCount(summarizer) === 6)
}
+ // TODO: this test should not be committed. It is here to isolate some
performance hotspots.
+ test("perf test") {
+val n = 1000
+val rdd1 = sc.parallelize(1 to n).map { idx =>
+ OldVectors.dense(idx.toDouble)
+}
+val trieouts = 10
--- End diff --
I did not try about that class, thanks. It should stabilize the results.
---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If your project does not have this feature
enabled and wishes so, or if the feature is enabled but not working, please
contact infrastructure at infrastruct...@apache.org or file a JIRA ticket
with INFRA.
---
-
To unsubscribe, e-mail: reviews-unsubscr...@spark.apache.org
For additional commands, e-mail: reviews-h...@spark.apache.org
[GitHub] spark pull request #17419: [SPARK-19634][ML] Multivariate summarizer - dataf...
Github user kiszk commented on a diff in the pull request:
https://github.com/apache/spark/pull/17419#discussion_r108742186
--- Diff:
mllib/src/test/scala/org/apache/spark/ml/stat/SummarizerSuite.scala ---
@@ -335,4 +335,65 @@ class SummarizerSuite extends SparkFunSuite with
MLlibTestSparkContext {
assert(Buffer.totalCount(summarizer) === 6)
}
+ // TODO: this test should not be committed. It is here to isolate some
performance hotspots.
+ test("perf test") {
+val n = 1000
+val rdd1 = sc.parallelize(1 to n).map { idx =>
+ OldVectors.dense(idx.toDouble)
+}
+val trieouts = 10
--- End diff --
I think that 10 times without warmup is too small for performance
measurement.
Can we use `Benchmark` class or add warmup run as `Benchmark` class does?
---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If your project does not have this feature
enabled and wishes so, or if the feature is enabled but not working, please
contact infrastructure at infrastruct...@apache.org or file a JIRA ticket
with INFRA.
---
-
To unsubscribe, e-mail: reviews-unsubscr...@spark.apache.org
For additional commands, e-mail: reviews-h...@spark.apache.org
[GitHub] spark pull request #17419: [SPARK-19634][ML] Multivariate summarizer - dataf...
Github user jkbradley commented on a diff in the pull request:
https://github.com/apache/spark/pull/17419#discussion_r108218440
--- Diff:
mllib/src/test/scala/org/apache/spark/ml/stat/SummarizerSuite.scala ---
@@ -0,0 +1,338 @@
+/*
+ * 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.
+ */
+
+package org.apache.spark.ml.stat
+
+import org.scalatest.exceptions.TestFailedException
+
+import org.apache.spark.{SparkException, SparkFunSuite}
+import org.apache.spark.ml.linalg.{Vector, Vectors}
+import org.apache.spark.ml.stat.SummaryBuilderImpl.Buffer
+import org.apache.spark.ml.util.TestingUtils._
+import org.apache.spark.mllib.linalg.{Vector => OldVector, Vectors =>
OldVectors}
+import org.apache.spark.mllib.stat.MultivariateOnlineSummarizer
+import org.apache.spark.mllib.util.MLlibTestSparkContext
+import org.apache.spark.sql.{DataFrame, Row}
+import org.apache.spark.sql.catalyst.expressions.GenericRowWithSchema
+
+class SummarizerSuite extends SparkFunSuite with MLlibTestSparkContext {
+
+ import testImplicits._
+ import Summarizer._
+
+ private case class ExpectedMetrics(
+ mean: Seq[Double],
+ variance: Seq[Double],
+ count: Long,
+ numNonZeros: Seq[Long],
+ max: Seq[Double],
+ min: Seq[Double],
+ normL2: Seq[Double],
+ normL1: Seq[Double])
+
+ // The input is expected to be either a sparse vector, a dense vector or
an array of doubles
+ // (which will be converted to a dense vector)
+ // The expected is the list of all the known metrics.
+ //
+ // The tests take an list of input vectors and a list of all the summary
values that
+ // are expected for this input. They currently test against some fixed
subset of the
+ // metrics, but should be made fuzzy in the future.
+
+ private def testExample(name: String, input: Seq[Any], exp:
ExpectedMetrics): Unit = {
+def inputVec: Seq[Vector] = input.map {
+ case x: Array[Double @unchecked] => Vectors.dense(x)
+ case x: Seq[Double @unchecked] => Vectors.dense(x.toArray)
+ case x: Vector => x
+ case x => throw new Exception(x.toString)
+}
+
+val s = {
--- End diff --
rename s -> summarizer
---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If your project does not have this feature
enabled and wishes so, or if the feature is enabled but not working, please
contact infrastructure at infrastruct...@apache.org or file a JIRA ticket
with INFRA.
---
-
To unsubscribe, e-mail: reviews-unsubscr...@spark.apache.org
For additional commands, e-mail: reviews-h...@spark.apache.org
[GitHub] spark pull request #17419: [SPARK-19634][ML] Multivariate summarizer - dataf...
Github user jkbradley commented on a diff in the pull request:
https://github.com/apache/spark/pull/17419#discussion_r108219483
--- Diff:
mllib/src/test/scala/org/apache/spark/ml/stat/SummarizerSuite.scala ---
@@ -0,0 +1,338 @@
+/*
+ * 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.
+ */
+
+package org.apache.spark.ml.stat
+
+import org.scalatest.exceptions.TestFailedException
+
+import org.apache.spark.{SparkException, SparkFunSuite}
+import org.apache.spark.ml.linalg.{Vector, Vectors}
+import org.apache.spark.ml.stat.SummaryBuilderImpl.Buffer
+import org.apache.spark.ml.util.TestingUtils._
+import org.apache.spark.mllib.linalg.{Vector => OldVector, Vectors =>
OldVectors}
+import org.apache.spark.mllib.stat.MultivariateOnlineSummarizer
+import org.apache.spark.mllib.util.MLlibTestSparkContext
+import org.apache.spark.sql.{DataFrame, Row}
+import org.apache.spark.sql.catalyst.expressions.GenericRowWithSchema
+
+class SummarizerSuite extends SparkFunSuite with MLlibTestSparkContext {
+
+ import testImplicits._
+ import Summarizer._
+
+ private case class ExpectedMetrics(
+ mean: Seq[Double],
+ variance: Seq[Double],
+ count: Long,
+ numNonZeros: Seq[Long],
+ max: Seq[Double],
+ min: Seq[Double],
+ normL2: Seq[Double],
+ normL1: Seq[Double])
+
+ // The input is expected to be either a sparse vector, a dense vector or
an array of doubles
+ // (which will be converted to a dense vector)
+ // The expected is the list of all the known metrics.
+ //
+ // The tests take an list of input vectors and a list of all the summary
values that
+ // are expected for this input. They currently test against some fixed
subset of the
+ // metrics, but should be made fuzzy in the future.
+
+ private def testExample(name: String, input: Seq[Any], exp:
ExpectedMetrics): Unit = {
+def inputVec: Seq[Vector] = input.map {
+ case x: Array[Double @unchecked] => Vectors.dense(x)
+ case x: Seq[Double @unchecked] => Vectors.dense(x.toArray)
+ case x: Vector => x
+ case x => throw new Exception(x.toString)
+}
+
+val s = {
+ val s2 = new MultivariateOnlineSummarizer
+ inputVec.foreach(v => s2.add(OldVectors.fromML(v)))
+ s2
+}
+
+// Because the Spark context is reset between tests, we cannot hold a
reference onto it.
+def wrapped() = {
+ val df = sc.parallelize(inputVec).map(Tuple1.apply).toDF("features")
+ val c = df.col("features")
+ (df, c)
+}
+
+registerTest(s"$name - mean only") {
+ val (df, c) = wrapped()
+ compare(df.select(metrics("mean").summary(c), mean(c)),
Seq(Row(exp.mean), s.mean))
+}
+
+registerTest(s"$name - mean only (direct)") {
+ val (df, c) = wrapped()
+ compare(df.select(mean(c)), Seq(exp.mean))
+}
+
+registerTest(s"$name - variance only") {
+ val (df, c) = wrapped()
+ compare(df.select(metrics("variance").summary(c), variance(c)),
+Seq(Row(exp.variance), s.variance))
+}
+
+registerTest(s"$name - variance only (direct)") {
+ val (df, c) = wrapped()
+ compare(df.select(variance(c)), Seq(s.variance))
+}
+
+registerTest(s"$name - count only") {
+ val (df, c) = wrapped()
+ compare(df.select(metrics("count").summary(c), count(c)),
+Seq(Row(exp.count), exp.count))
+}
+
+registerTest(s"$name - count only (direct)") {
+ val (df, c) = wrapped()
+ compare(df.select(count(c)),
+Seq(exp.count))
+}
+
+registerTest(s"$name - numNonZeros only") {
+ val (df, c) = wrapped()
+ compare(df.select(metrics("numNonZeros").summary(c), numNonZeros(c)),
+Seq(Row(exp.numNonZeros), exp.numNonZeros))
+}
+
+registerTest(s"$name - numNonZeros only
[GitHub] spark pull request #17419: [SPARK-19634][ML] Multivariate summarizer - dataf...
Github user jkbradley commented on a diff in the pull request:
https://github.com/apache/spark/pull/17419#discussion_r108216365
--- Diff:
mllib/src/test/scala/org/apache/spark/ml/stat/SummarizerSuite.scala ---
@@ -0,0 +1,338 @@
+/*
+ * 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.
+ */
+
+package org.apache.spark.ml.stat
+
+import org.scalatest.exceptions.TestFailedException
+
+import org.apache.spark.{SparkException, SparkFunSuite}
+import org.apache.spark.ml.linalg.{Vector, Vectors}
+import org.apache.spark.ml.stat.SummaryBuilderImpl.Buffer
+import org.apache.spark.ml.util.TestingUtils._
+import org.apache.spark.mllib.linalg.{Vector => OldVector, Vectors =>
OldVectors}
+import org.apache.spark.mllib.stat.MultivariateOnlineSummarizer
+import org.apache.spark.mllib.util.MLlibTestSparkContext
+import org.apache.spark.sql.{DataFrame, Row}
+import org.apache.spark.sql.catalyst.expressions.GenericRowWithSchema
+
+class SummarizerSuite extends SparkFunSuite with MLlibTestSparkContext {
+
+ import testImplicits._
+ import Summarizer._
+
+ private case class ExpectedMetrics(
+ mean: Seq[Double],
+ variance: Seq[Double],
+ count: Long,
+ numNonZeros: Seq[Long],
+ max: Seq[Double],
+ min: Seq[Double],
+ normL2: Seq[Double],
+ normL1: Seq[Double])
+
+ // The input is expected to be either a sparse vector, a dense vector or
an array of doubles
+ // (which will be converted to a dense vector)
+ // The expected is the list of all the known metrics.
+ //
+ // The tests take an list of input vectors and a list of all the summary
values that
+ // are expected for this input. They currently test against some fixed
subset of the
+ // metrics, but should be made fuzzy in the future.
--- End diff --
fuzzy?
---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If your project does not have this feature
enabled and wishes so, or if the feature is enabled but not working, please
contact infrastructure at infrastruct...@apache.org or file a JIRA ticket
with INFRA.
---
-
To unsubscribe, e-mail: reviews-unsubscr...@spark.apache.org
For additional commands, e-mail: reviews-h...@spark.apache.org
[GitHub] spark pull request #17419: [SPARK-19634][ML] Multivariate summarizer - dataf...
Github user jkbradley commented on a diff in the pull request:
https://github.com/apache/spark/pull/17419#discussion_r108215942
--- Diff: mllib/src/main/scala/org/apache/spark/ml/stat/Summarizer.scala ---
@@ -0,0 +1,746 @@
+/*
+ * 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.
+ */
+
+package org.apache.spark.ml.stat
+
+import breeze.{linalg => la}
+import breeze.linalg.{Vector => BV}
+import breeze.numerics
+
+import org.apache.spark.SparkException
+import org.apache.spark.annotation.Since
+import org.apache.spark.internal.Logging
+import org.apache.spark.ml.linalg.{DenseVector, SparseVector, Vector,
Vectors, VectorUDT}
+import org.apache.spark.sql.Column
+import org.apache.spark.sql.catalyst.InternalRow
+import org.apache.spark.sql.catalyst.expressions.{Expression,
UnsafeArrayData, UnsafeProjection, UnsafeRow}
+import
org.apache.spark.sql.catalyst.expressions.aggregate.{AggregateExpression,
Complete, TypedImperativeAggregate}
+import org.apache.spark.sql.types._
+
+
+/**
+ * A builder object that provides summary statistics about a given column.
+ *
+ * Users should not directly create such builders, but instead use one of
the methods in
+ * [[Summarizer]].
+ */
+@Since("2.2.0")
+abstract class SummaryBuilder {
+ /**
+ * Returns an aggregate object that contains the summary of the column
with the requested metrics.
+ * @param column a column that contains Vector object.
+ * @return an aggregate column that contains the statistics. The exact
content of this
+ * structure is determined during the creation of the builder.
+ */
+ @Since("2.2.0")
+ def summary(column: Column): Column
+}
+
+/**
+ * Tools for vectorized statistics on MLlib Vectors.
+ *
+ * The methods in this package provide various statistics for Vectors
contained inside DataFrames.
+ *
+ * This class lets users pick the statistics they would like to extract
for a given column. Here is
+ * an example in Scala:
+ * {{{
+ * val dataframe = ... // Some dataframe containing a feature column
+ * val allStats = dataframe.select(Summarizer.metrics("min",
"max").summary($"features"))
+ * val Row(min_, max_) = allStats.first()
+ * }}}
+ *
+ * If one wants to get a single metric, shortcuts are also available:
+ * {{{
+ * val meanDF = dataframe.select(Summarizer.mean($"features"))
+ * val Row(mean_) = meanDF.first()
+ * }}}
+ */
+@Since("2.2.0")
+object Summarizer extends Logging {
--- End diff --
Mark Experimental
---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If your project does not have this feature
enabled and wishes so, or if the feature is enabled but not working, please
contact infrastructure at infrastruct...@apache.org or file a JIRA ticket
with INFRA.
---
-
To unsubscribe, e-mail: reviews-unsubscr...@spark.apache.org
For additional commands, e-mail: reviews-h...@spark.apache.org
[GitHub] spark pull request #17419: [SPARK-19634][ML] Multivariate summarizer - dataf...
Github user jkbradley commented on a diff in the pull request:
https://github.com/apache/spark/pull/17419#discussion_r108218508
--- Diff:
mllib/src/test/scala/org/apache/spark/ml/stat/SummarizerSuite.scala ---
@@ -0,0 +1,338 @@
+/*
+ * 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.
+ */
+
+package org.apache.spark.ml.stat
+
+import org.scalatest.exceptions.TestFailedException
+
+import org.apache.spark.{SparkException, SparkFunSuite}
+import org.apache.spark.ml.linalg.{Vector, Vectors}
+import org.apache.spark.ml.stat.SummaryBuilderImpl.Buffer
+import org.apache.spark.ml.util.TestingUtils._
+import org.apache.spark.mllib.linalg.{Vector => OldVector, Vectors =>
OldVectors}
+import org.apache.spark.mllib.stat.MultivariateOnlineSummarizer
+import org.apache.spark.mllib.util.MLlibTestSparkContext
+import org.apache.spark.sql.{DataFrame, Row}
+import org.apache.spark.sql.catalyst.expressions.GenericRowWithSchema
+
+class SummarizerSuite extends SparkFunSuite with MLlibTestSparkContext {
+
+ import testImplicits._
+ import Summarizer._
+
+ private case class ExpectedMetrics(
+ mean: Seq[Double],
+ variance: Seq[Double],
+ count: Long,
+ numNonZeros: Seq[Long],
+ max: Seq[Double],
+ min: Seq[Double],
+ normL2: Seq[Double],
+ normL1: Seq[Double])
+
+ // The input is expected to be either a sparse vector, a dense vector or
an array of doubles
+ // (which will be converted to a dense vector)
+ // The expected is the list of all the known metrics.
+ //
+ // The tests take an list of input vectors and a list of all the summary
values that
+ // are expected for this input. They currently test against some fixed
subset of the
+ // metrics, but should be made fuzzy in the future.
+
+ private def testExample(name: String, input: Seq[Any], exp:
ExpectedMetrics): Unit = {
+def inputVec: Seq[Vector] = input.map {
+ case x: Array[Double @unchecked] => Vectors.dense(x)
+ case x: Seq[Double @unchecked] => Vectors.dense(x.toArray)
+ case x: Vector => x
+ case x => throw new Exception(x.toString)
+}
+
+val s = {
+ val s2 = new MultivariateOnlineSummarizer
+ inputVec.foreach(v => s2.add(OldVectors.fromML(v)))
+ s2
+}
+
+// Because the Spark context is reset between tests, we cannot hold a
reference onto it.
+def wrapped() = {
--- End diff --
rename wrapped -> wrappedInit
---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If your project does not have this feature
enabled and wishes so, or if the feature is enabled but not working, please
contact infrastructure at infrastruct...@apache.org or file a JIRA ticket
with INFRA.
---
-
To unsubscribe, e-mail: reviews-unsubscr...@spark.apache.org
For additional commands, e-mail: reviews-h...@spark.apache.org
[GitHub] spark pull request #17419: [SPARK-19634][ML] Multivariate summarizer - dataf...
Github user jkbradley commented on a diff in the pull request:
https://github.com/apache/spark/pull/17419#discussion_r108216014
--- Diff: mllib/src/main/scala/org/apache/spark/ml/stat/Summarizer.scala ---
@@ -0,0 +1,746 @@
+/*
+ * 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.
+ */
+
+package org.apache.spark.ml.stat
+
+import breeze.{linalg => la}
+import breeze.linalg.{Vector => BV}
+import breeze.numerics
+
+import org.apache.spark.SparkException
+import org.apache.spark.annotation.Since
+import org.apache.spark.internal.Logging
+import org.apache.spark.ml.linalg.{DenseVector, SparseVector, Vector,
Vectors, VectorUDT}
+import org.apache.spark.sql.Column
+import org.apache.spark.sql.catalyst.InternalRow
+import org.apache.spark.sql.catalyst.expressions.{Expression,
UnsafeArrayData, UnsafeProjection, UnsafeRow}
+import
org.apache.spark.sql.catalyst.expressions.aggregate.{AggregateExpression,
Complete, TypedImperativeAggregate}
+import org.apache.spark.sql.types._
+
+
+/**
+ * A builder object that provides summary statistics about a given column.
+ *
+ * Users should not directly create such builders, but instead use one of
the methods in
+ * [[Summarizer]].
+ */
+@Since("2.2.0")
+abstract class SummaryBuilder {
+ /**
+ * Returns an aggregate object that contains the summary of the column
with the requested metrics.
+ * @param column a column that contains Vector object.
+ * @return an aggregate column that contains the statistics. The exact
content of this
+ * structure is determined during the creation of the builder.
+ */
+ @Since("2.2.0")
+ def summary(column: Column): Column
+}
+
+/**
+ * Tools for vectorized statistics on MLlib Vectors.
+ *
+ * The methods in this package provide various statistics for Vectors
contained inside DataFrames.
+ *
+ * This class lets users pick the statistics they would like to extract
for a given column. Here is
+ * an example in Scala:
+ * {{{
+ * val dataframe = ... // Some dataframe containing a feature column
+ * val allStats = dataframe.select(Summarizer.metrics("min",
"max").summary($"features"))
+ * val Row(min_, max_) = allStats.first()
+ * }}}
+ *
+ * If one wants to get a single metric, shortcuts are also available:
+ * {{{
+ * val meanDF = dataframe.select(Summarizer.mean($"features"))
+ * val Row(mean_) = meanDF.first()
+ * }}}
+ */
+@Since("2.2.0")
+object Summarizer extends Logging {
+
+ import SummaryBuilderImpl._
+
+ /**
+ * Given a list of metrics, provides a builder that it turns computes
metrics from a column.
+ *
+ * See the documentation of [[Summarizer]] for an example.
+ *
+ * The following metrics are accepted (case sensitive):
+ * - mean: a vector that contains the coefficient-wise mean.
+ * - variance: a vector tha contains the coefficient-wise variance.
+ * - count: the count of all vectors seen.
+ * - numNonzeros: a vector with the number of non-zeros for each
coefficients
+ * - max: the maximum for each coefficient.
+ * - min: the minimum for each coefficient.
+ * - normL2: the Euclidian norm for each coefficient.
+ * - normL1: the L1 norm of each coefficient (sum of the absolute
values).
+ * @param firstMetric the metric being provided
+ * @param metrics additional metrics that can be provided.
+ * @return a builder.
+ * @throws IllegalArgumentException if one of the metric names is not
understood.
+ */
+ @Since("2.2.0")
+ def metrics(firstMetric: String, metrics: String*): SummaryBuilder = {
+val (typedMetrics, computeMetrics) =
getRelevantMetrics(Seq(firstMetric) ++ metrics)
+new SummaryBuilderImpl(typedMetrics, computeMetrics)
+ }
+
+ def mean(col: Column): Column = getSingleMetric(col, "mean")
--- End diff --
Add Since annotations for all of these public methods
[GitHub] spark pull request #17419: [SPARK-19634][ML] Multivariate summarizer - dataf...
Github user jkbradley commented on a diff in the pull request:
https://github.com/apache/spark/pull/17419#discussion_r108219172
--- Diff:
mllib/src/test/scala/org/apache/spark/ml/stat/SummarizerSuite.scala ---
@@ -0,0 +1,338 @@
+/*
+ * 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.
+ */
+
+package org.apache.spark.ml.stat
+
+import org.scalatest.exceptions.TestFailedException
+
+import org.apache.spark.{SparkException, SparkFunSuite}
+import org.apache.spark.ml.linalg.{Vector, Vectors}
+import org.apache.spark.ml.stat.SummaryBuilderImpl.Buffer
+import org.apache.spark.ml.util.TestingUtils._
+import org.apache.spark.mllib.linalg.{Vector => OldVector, Vectors =>
OldVectors}
+import org.apache.spark.mllib.stat.MultivariateOnlineSummarizer
+import org.apache.spark.mllib.util.MLlibTestSparkContext
+import org.apache.spark.sql.{DataFrame, Row}
+import org.apache.spark.sql.catalyst.expressions.GenericRowWithSchema
+
+class SummarizerSuite extends SparkFunSuite with MLlibTestSparkContext {
+
+ import testImplicits._
+ import Summarizer._
+
+ private case class ExpectedMetrics(
+ mean: Seq[Double],
+ variance: Seq[Double],
+ count: Long,
+ numNonZeros: Seq[Long],
+ max: Seq[Double],
+ min: Seq[Double],
+ normL2: Seq[Double],
+ normL1: Seq[Double])
+
+ // The input is expected to be either a sparse vector, a dense vector or
an array of doubles
+ // (which will be converted to a dense vector)
+ // The expected is the list of all the known metrics.
+ //
+ // The tests take an list of input vectors and a list of all the summary
values that
+ // are expected for this input. They currently test against some fixed
subset of the
+ // metrics, but should be made fuzzy in the future.
+
+ private def testExample(name: String, input: Seq[Any], exp:
ExpectedMetrics): Unit = {
+def inputVec: Seq[Vector] = input.map {
+ case x: Array[Double @unchecked] => Vectors.dense(x)
+ case x: Seq[Double @unchecked] => Vectors.dense(x.toArray)
+ case x: Vector => x
+ case x => throw new Exception(x.toString)
+}
+
+val s = {
+ val s2 = new MultivariateOnlineSummarizer
+ inputVec.foreach(v => s2.add(OldVectors.fromML(v)))
+ s2
+}
+
+// Because the Spark context is reset between tests, we cannot hold a
reference onto it.
+def wrapped() = {
+ val df = sc.parallelize(inputVec).map(Tuple1.apply).toDF("features")
+ val c = df.col("features")
+ (df, c)
+}
+
+registerTest(s"$name - mean only") {
+ val (df, c) = wrapped()
+ compare(df.select(metrics("mean").summary(c), mean(c)),
Seq(Row(exp.mean), s.mean))
+}
+
+registerTest(s"$name - mean only (direct)") {
+ val (df, c) = wrapped()
+ compare(df.select(mean(c)), Seq(exp.mean))
+}
+
+registerTest(s"$name - variance only") {
+ val (df, c) = wrapped()
+ compare(df.select(metrics("variance").summary(c), variance(c)),
+Seq(Row(exp.variance), s.variance))
+}
+
+registerTest(s"$name - variance only (direct)") {
+ val (df, c) = wrapped()
+ compare(df.select(variance(c)), Seq(s.variance))
+}
+
+registerTest(s"$name - count only") {
+ val (df, c) = wrapped()
+ compare(df.select(metrics("count").summary(c), count(c)),
+Seq(Row(exp.count), exp.count))
+}
+
+registerTest(s"$name - count only (direct)") {
+ val (df, c) = wrapped()
+ compare(df.select(count(c)),
+Seq(exp.count))
+}
+
+registerTest(s"$name - numNonZeros only") {
+ val (df, c) = wrapped()
+ compare(df.select(metrics("numNonZeros").summary(c), numNonZeros(c)),
+Seq(Row(exp.numNonZeros), exp.numNonZeros))
+}
+
+registerTest(s"$name - numNonZeros only
[GitHub] spark pull request #17419: [SPARK-19634][ML] Multivariate summarizer - dataf...
Github user jkbradley commented on a diff in the pull request:
https://github.com/apache/spark/pull/17419#discussion_r108219945
--- Diff:
mllib/src/test/scala/org/apache/spark/ml/stat/SummarizerSuite.scala ---
@@ -0,0 +1,338 @@
+/*
+ * 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.
+ */
+
+package org.apache.spark.ml.stat
+
+import org.scalatest.exceptions.TestFailedException
+
+import org.apache.spark.{SparkException, SparkFunSuite}
+import org.apache.spark.ml.linalg.{Vector, Vectors}
+import org.apache.spark.ml.stat.SummaryBuilderImpl.Buffer
+import org.apache.spark.ml.util.TestingUtils._
+import org.apache.spark.mllib.linalg.{Vector => OldVector, Vectors =>
OldVectors}
+import org.apache.spark.mllib.stat.MultivariateOnlineSummarizer
+import org.apache.spark.mllib.util.MLlibTestSparkContext
+import org.apache.spark.sql.{DataFrame, Row}
+import org.apache.spark.sql.catalyst.expressions.GenericRowWithSchema
+
+class SummarizerSuite extends SparkFunSuite with MLlibTestSparkContext {
+
+ import testImplicits._
+ import Summarizer._
+
+ private case class ExpectedMetrics(
+ mean: Seq[Double],
+ variance: Seq[Double],
+ count: Long,
+ numNonZeros: Seq[Long],
+ max: Seq[Double],
+ min: Seq[Double],
+ normL2: Seq[Double],
+ normL1: Seq[Double])
+
+ // The input is expected to be either a sparse vector, a dense vector or
an array of doubles
+ // (which will be converted to a dense vector)
+ // The expected is the list of all the known metrics.
+ //
+ // The tests take an list of input vectors and a list of all the summary
values that
+ // are expected for this input. They currently test against some fixed
subset of the
+ // metrics, but should be made fuzzy in the future.
+
+ private def testExample(name: String, input: Seq[Any], exp:
ExpectedMetrics): Unit = {
+def inputVec: Seq[Vector] = input.map {
+ case x: Array[Double @unchecked] => Vectors.dense(x)
+ case x: Seq[Double @unchecked] => Vectors.dense(x.toArray)
+ case x: Vector => x
+ case x => throw new Exception(x.toString)
+}
+
+val s = {
+ val s2 = new MultivariateOnlineSummarizer
+ inputVec.foreach(v => s2.add(OldVectors.fromML(v)))
+ s2
+}
+
+// Because the Spark context is reset between tests, we cannot hold a
reference onto it.
+def wrapped() = {
+ val df = sc.parallelize(inputVec).map(Tuple1.apply).toDF("features")
+ val c = df.col("features")
+ (df, c)
+}
+
+registerTest(s"$name - mean only") {
+ val (df, c) = wrapped()
+ compare(df.select(metrics("mean").summary(c), mean(c)),
Seq(Row(exp.mean), s.mean))
+}
+
+registerTest(s"$name - mean only (direct)") {
+ val (df, c) = wrapped()
+ compare(df.select(mean(c)), Seq(exp.mean))
+}
+
+registerTest(s"$name - variance only") {
+ val (df, c) = wrapped()
+ compare(df.select(metrics("variance").summary(c), variance(c)),
+Seq(Row(exp.variance), s.variance))
+}
+
+registerTest(s"$name - variance only (direct)") {
+ val (df, c) = wrapped()
+ compare(df.select(variance(c)), Seq(s.variance))
+}
+
+registerTest(s"$name - count only") {
+ val (df, c) = wrapped()
+ compare(df.select(metrics("count").summary(c), count(c)),
+Seq(Row(exp.count), exp.count))
+}
+
+registerTest(s"$name - count only (direct)") {
+ val (df, c) = wrapped()
+ compare(df.select(count(c)),
+Seq(exp.count))
+}
+
+registerTest(s"$name - numNonZeros only") {
+ val (df, c) = wrapped()
+ compare(df.select(metrics("numNonZeros").summary(c), numNonZeros(c)),
+Seq(Row(exp.numNonZeros), exp.numNonZeros))
+}
+
+registerTest(s"$name - numNonZeros only
[GitHub] spark pull request #17419: [SPARK-19634][ML] Multivariate summarizer - dataf...
Github user jkbradley commented on a diff in the pull request:
https://github.com/apache/spark/pull/17419#discussion_r108220085
--- Diff:
mllib/src/test/scala/org/apache/spark/ml/stat/SummarizerSuite.scala ---
@@ -0,0 +1,338 @@
+/*
+ * 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.
+ */
+
+package org.apache.spark.ml.stat
+
+import org.scalatest.exceptions.TestFailedException
+
+import org.apache.spark.{SparkException, SparkFunSuite}
+import org.apache.spark.ml.linalg.{Vector, Vectors}
+import org.apache.spark.ml.stat.SummaryBuilderImpl.Buffer
+import org.apache.spark.ml.util.TestingUtils._
+import org.apache.spark.mllib.linalg.{Vector => OldVector, Vectors =>
OldVectors}
+import org.apache.spark.mllib.stat.MultivariateOnlineSummarizer
+import org.apache.spark.mllib.util.MLlibTestSparkContext
+import org.apache.spark.sql.{DataFrame, Row}
+import org.apache.spark.sql.catalyst.expressions.GenericRowWithSchema
+
+class SummarizerSuite extends SparkFunSuite with MLlibTestSparkContext {
+
+ import testImplicits._
+ import Summarizer._
+
+ private case class ExpectedMetrics(
+ mean: Seq[Double],
+ variance: Seq[Double],
+ count: Long,
+ numNonZeros: Seq[Long],
+ max: Seq[Double],
+ min: Seq[Double],
+ normL2: Seq[Double],
+ normL1: Seq[Double])
+
+ // The input is expected to be either a sparse vector, a dense vector or
an array of doubles
+ // (which will be converted to a dense vector)
+ // The expected is the list of all the known metrics.
+ //
+ // The tests take an list of input vectors and a list of all the summary
values that
+ // are expected for this input. They currently test against some fixed
subset of the
+ // metrics, but should be made fuzzy in the future.
+
+ private def testExample(name: String, input: Seq[Any], exp:
ExpectedMetrics): Unit = {
+def inputVec: Seq[Vector] = input.map {
+ case x: Array[Double @unchecked] => Vectors.dense(x)
+ case x: Seq[Double @unchecked] => Vectors.dense(x.toArray)
+ case x: Vector => x
+ case x => throw new Exception(x.toString)
+}
+
+val s = {
+ val s2 = new MultivariateOnlineSummarizer
+ inputVec.foreach(v => s2.add(OldVectors.fromML(v)))
+ s2
+}
+
+// Because the Spark context is reset between tests, we cannot hold a
reference onto it.
+def wrapped() = {
+ val df = sc.parallelize(inputVec).map(Tuple1.apply).toDF("features")
+ val c = df.col("features")
+ (df, c)
+}
+
+registerTest(s"$name - mean only") {
+ val (df, c) = wrapped()
+ compare(df.select(metrics("mean").summary(c), mean(c)),
Seq(Row(exp.mean), s.mean))
+}
+
+registerTest(s"$name - mean only (direct)") {
+ val (df, c) = wrapped()
+ compare(df.select(mean(c)), Seq(exp.mean))
+}
+
+registerTest(s"$name - variance only") {
+ val (df, c) = wrapped()
+ compare(df.select(metrics("variance").summary(c), variance(c)),
+Seq(Row(exp.variance), s.variance))
+}
+
+registerTest(s"$name - variance only (direct)") {
+ val (df, c) = wrapped()
+ compare(df.select(variance(c)), Seq(s.variance))
+}
+
+registerTest(s"$name - count only") {
+ val (df, c) = wrapped()
+ compare(df.select(metrics("count").summary(c), count(c)),
+Seq(Row(exp.count), exp.count))
+}
+
+registerTest(s"$name - count only (direct)") {
+ val (df, c) = wrapped()
+ compare(df.select(count(c)),
+Seq(exp.count))
+}
+
+registerTest(s"$name - numNonZeros only") {
+ val (df, c) = wrapped()
+ compare(df.select(metrics("numNonZeros").summary(c), numNonZeros(c)),
+Seq(Row(exp.numNonZeros), exp.numNonZeros))
+}
+
+registerTest(s"$name - numNonZeros only
[GitHub] spark pull request #17419: [SPARK-19634][ML] Multivariate summarizer - dataf...
Github user jkbradley commented on a diff in the pull request:
https://github.com/apache/spark/pull/17419#discussion_r108215935
--- Diff: mllib/src/main/scala/org/apache/spark/ml/stat/Summarizer.scala ---
@@ -0,0 +1,746 @@
+/*
+ * 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.
+ */
+
+package org.apache.spark.ml.stat
+
+import breeze.{linalg => la}
+import breeze.linalg.{Vector => BV}
+import breeze.numerics
+
+import org.apache.spark.SparkException
+import org.apache.spark.annotation.Since
+import org.apache.spark.internal.Logging
+import org.apache.spark.ml.linalg.{DenseVector, SparseVector, Vector,
Vectors, VectorUDT}
+import org.apache.spark.sql.Column
+import org.apache.spark.sql.catalyst.InternalRow
+import org.apache.spark.sql.catalyst.expressions.{Expression,
UnsafeArrayData, UnsafeProjection, UnsafeRow}
+import
org.apache.spark.sql.catalyst.expressions.aggregate.{AggregateExpression,
Complete, TypedImperativeAggregate}
+import org.apache.spark.sql.types._
+
+
+/**
+ * A builder object that provides summary statistics about a given column.
+ *
+ * Users should not directly create such builders, but instead use one of
the methods in
+ * [[Summarizer]].
+ */
+@Since("2.2.0")
+abstract class SummaryBuilder {
--- End diff --
Make this sealed
Also mark Experimental
---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If your project does not have this feature
enabled and wishes so, or if the feature is enabled but not working, please
contact infrastructure at infrastruct...@apache.org or file a JIRA ticket
with INFRA.
---
-
To unsubscribe, e-mail: reviews-unsubscr...@spark.apache.org
For additional commands, e-mail: reviews-h...@spark.apache.org
