[GitHub] [spark] zhengruifeng commented on a change in pull request #27623: [SPARK-30867][ML] Add FValueRegressionTest
zhengruifeng commented on a change in pull request #27623: [SPARK-30867][ML]
Add FValueRegressionTest
URL: https://github.com/apache/spark/pull/27623#discussion_r381063994
##
File path: mllib/src/main/scala/org/apache/spark/ml/stat/SelectionTest.scala
##
@@ -0,0 +1,100 @@
+/*
+ * 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.apache.commons.math3.distribution.FDistribution
+
+import org.apache.spark.annotation.Since
+import org.apache.spark.ml.feature.LabeledPoint
+import org.apache.spark.ml.linalg.{Vector, VectorUDT}
+import org.apache.spark.ml.util.SchemaUtils
+import org.apache.spark.sql.{Dataset, Row}
+import org.apache.spark.sql.functions.{avg, col, stddev}
+
+@Since("3.1.0")
+object SelectionTest {
+
+ /**
+ * @param dataset DataFrame of continuous labels and continuous features.
+ * @param featuresCol Name of features column in dataset, of type `Vector`
(`VectorUDT`)
+ * @param labelCol Name of label column in dataset, of any numerical type
+ * @return Array containing the SelectionTestResult for every feature
against the label.
+ */
+ @Since("3.1.0")
+ def fValueRegressionTest(dataset: Dataset[_], featuresCol: String, labelCol:
String):
+Array[SelectionTestResult] = {
+
+val spark = dataset.sparkSession
+import spark.implicits._
+
+SchemaUtils.checkColumnType(dataset.schema, featuresCol, new VectorUDT)
+SchemaUtils.checkNumericType(dataset.schema, labelCol)
+
+val Row(xMeans: Vector, xStd: Vector, yMean: Double, yStd: Double, count:
Long) = dataset
+ .select(Summarizer.metrics("mean", "std",
"count").summary(col(featuresCol)).as("summary"),
+avg(col(labelCol)).as("yMean"),
+stddev(col(labelCol)).as("yStd"))
+ .select("summary.mean", "summary.std", "yMean", "yStd", "summary.count")
+ .first()
+
+val labeledPointRdd = dataset.select(col("label").cast("double"),
col("features"))
+ .as[(Double, Vector)]
+ .rdd.map { case (label, features) => LabeledPoint(label, features) }
+
+val numFeatures = labeledPointRdd.first().features.size
Review comment:
using `val numFeatures = xMeans.size` instead to avoid this `first` job
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org
With regards,
Apache Git Services
-
To unsubscribe, e-mail: reviews-unsubscr...@spark.apache.org
For additional commands, e-mail: reviews-h...@spark.apache.org
[GitHub] [spark] zhengruifeng commented on a change in pull request #27623: [SPARK-30867][ML] Add FValueRegressionTest
zhengruifeng commented on a change in pull request #27623: [SPARK-30867][ML]
Add FValueRegressionTest
URL: https://github.com/apache/spark/pull/27623#discussion_r381068169
##
File path:
mllib/src/main/scala/org/apache/spark/ml/stat/SelectionTestResult.scala
##
@@ -0,0 +1,85 @@
+/*
+ * 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.apache.spark.annotation.Since
+
+/**
+ * Trait for selection test results.
+ */
+@Since("3.1.0")
+trait SelectionTestResult {
Review comment:
`Selector` should be in `.feature` like `ChiSqSelector`
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org
With regards,
Apache Git Services
-
To unsubscribe, e-mail: reviews-unsubscr...@spark.apache.org
For additional commands, e-mail: reviews-h...@spark.apache.org
[GitHub] [spark] zhengruifeng commented on a change in pull request #27623: [SPARK-30867][ML] Add FValueRegressionTest
zhengruifeng commented on a change in pull request #27623: [SPARK-30867][ML]
Add FValueRegressionTest
URL: https://github.com/apache/spark/pull/27623#discussion_r381064590
##
File path: mllib/src/main/scala/org/apache/spark/ml/stat/SelectionTest.scala
##
@@ -0,0 +1,100 @@
+/*
+ * 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.apache.commons.math3.distribution.FDistribution
+
+import org.apache.spark.annotation.Since
+import org.apache.spark.ml.feature.LabeledPoint
+import org.apache.spark.ml.linalg.{Vector, VectorUDT}
+import org.apache.spark.ml.util.SchemaUtils
+import org.apache.spark.sql.{Dataset, Row}
+import org.apache.spark.sql.functions.{avg, col, stddev}
+
+@Since("3.1.0")
+object SelectionTest {
+
+ /**
+ * @param dataset DataFrame of continuous labels and continuous features.
+ * @param featuresCol Name of features column in dataset, of type `Vector`
(`VectorUDT`)
+ * @param labelCol Name of label column in dataset, of any numerical type
+ * @return Array containing the SelectionTestResult for every feature
against the label.
+ */
+ @Since("3.1.0")
+ def fValueRegressionTest(dataset: Dataset[_], featuresCol: String, labelCol:
String):
+Array[SelectionTestResult] = {
+
+val spark = dataset.sparkSession
+import spark.implicits._
+
+SchemaUtils.checkColumnType(dataset.schema, featuresCol, new VectorUDT)
+SchemaUtils.checkNumericType(dataset.schema, labelCol)
+
+val Row(xMeans: Vector, xStd: Vector, yMean: Double, yStd: Double, count:
Long) = dataset
+ .select(Summarizer.metrics("mean", "std",
"count").summary(col(featuresCol)).as("summary"),
+avg(col(labelCol)).as("yMean"),
+stddev(col(labelCol)).as("yStd"))
+ .select("summary.mean", "summary.std", "yMean", "yStd", "summary.count")
+ .first()
+
+val labeledPointRdd = dataset.select(col("label").cast("double"),
col("features"))
+ .as[(Double, Vector)]
+ .rdd.map { case (label, features) => LabeledPoint(label, features) }
+
+val numFeatures = labeledPointRdd.first().features.size
+val numSamples = count
+val degreeOfFreedom = numSamples.toInt - 2
+var fTestResultArray = new Array[SelectionTestResult](numFeatures)
+
+// Use two pass equation Cov[X,Y] = E[(X - E[X]) * (Y - E[Y])] to compute
covariance because
+// one pass equation Cov[X,Y] = E[XY] - E[X]E[Y] is susceptible to
catastrophic cancellation
+//
Review comment:
nit: remove this line?
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org
With regards,
Apache Git Services
-
To unsubscribe, e-mail: reviews-unsubscr...@spark.apache.org
For additional commands, e-mail: reviews-h...@spark.apache.org
[GitHub] [spark] zhengruifeng commented on a change in pull request #27623: [SPARK-30867][ML] Add FValueRegressionTest
zhengruifeng commented on a change in pull request #27623: [SPARK-30867][ML]
Add FValueRegressionTest
URL: https://github.com/apache/spark/pull/27623#discussion_r381065990
##
File path: mllib/src/main/scala/org/apache/spark/ml/stat/SelectionTest.scala
##
@@ -0,0 +1,100 @@
+/*
+ * 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.apache.commons.math3.distribution.FDistribution
+
+import org.apache.spark.annotation.Since
+import org.apache.spark.ml.feature.LabeledPoint
+import org.apache.spark.ml.linalg.{Vector, VectorUDT}
+import org.apache.spark.ml.util.SchemaUtils
+import org.apache.spark.sql.{Dataset, Row}
+import org.apache.spark.sql.functions.{avg, col, stddev}
+
+@Since("3.1.0")
+object SelectionTest {
Review comment:
Why adding such a class and file? It seems that current `XXXTest` are in
spearate files. Like `ChiSquareTest`, `KolmogorovSmirnovTest`.
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org
With regards,
Apache Git Services
-
To unsubscribe, e-mail: reviews-unsubscr...@spark.apache.org
For additional commands, e-mail: reviews-h...@spark.apache.org
[GitHub] [spark] zhengruifeng commented on a change in pull request #27623: [SPARK-30867][ML] Add FValueRegressionTest
zhengruifeng commented on a change in pull request #27623: [SPARK-30867][ML]
Add FValueRegressionTest
URL: https://github.com/apache/spark/pull/27623#discussion_r381067119
##
File path: mllib/src/main/scala/org/apache/spark/ml/stat/SelectionTest.scala
##
@@ -0,0 +1,100 @@
+/*
+ * 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.apache.commons.math3.distribution.FDistribution
+
+import org.apache.spark.annotation.Since
+import org.apache.spark.ml.feature.LabeledPoint
+import org.apache.spark.ml.linalg.{Vector, VectorUDT}
+import org.apache.spark.ml.util.SchemaUtils
+import org.apache.spark.sql.{Dataset, Row}
+import org.apache.spark.sql.functions.{avg, col, stddev}
+
+@Since("3.1.0")
+object SelectionTest {
+
+ /**
+ * @param dataset DataFrame of continuous labels and continuous features.
+ * @param featuresCol Name of features column in dataset, of type `Vector`
(`VectorUDT`)
+ * @param labelCol Name of label column in dataset, of any numerical type
+ * @return Array containing the SelectionTestResult for every feature
against the label.
+ */
+ @Since("3.1.0")
+ def fValueRegressionTest(dataset: Dataset[_], featuresCol: String, labelCol:
String):
+Array[SelectionTestResult] = {
+
+val spark = dataset.sparkSession
+import spark.implicits._
+
+SchemaUtils.checkColumnType(dataset.schema, featuresCol, new VectorUDT)
+SchemaUtils.checkNumericType(dataset.schema, labelCol)
+
+val Row(xMeans: Vector, xStd: Vector, yMean: Double, yStd: Double, count:
Long) = dataset
+ .select(Summarizer.metrics("mean", "std",
"count").summary(col(featuresCol)).as("summary"),
+avg(col(labelCol)).as("yMean"),
+stddev(col(labelCol)).as("yStd"))
+ .select("summary.mean", "summary.std", "yMean", "yStd", "summary.count")
+ .first()
+
+val labeledPointRdd = dataset.select(col("label").cast("double"),
col("features"))
+ .as[(Double, Vector)]
+ .rdd.map { case (label, features) => LabeledPoint(label, features) }
+
+val numFeatures = labeledPointRdd.first().features.size
+val numSamples = count
+val degreeOfFreedom = numSamples.toInt - 2
+var fTestResultArray = new Array[SelectionTestResult](numFeatures)
+
+// Use two pass equation Cov[X,Y] = E[(X - E[X]) * (Y - E[Y])] to compute
covariance because
+// one pass equation Cov[X,Y] = E[XY] - E[X]E[Y] is susceptible to
catastrophic cancellation
+//
+// sumForCov = Sum(((Xi - Avg(X)) * ((Yi-Avg(Y)))
+val sumForCov = labeledPointRdd.mapPartitions { iter =>
+ if (iter.hasNext) {
+val array = Array.ofDim[Double](numFeatures)
+while(iter.hasNext) {
+ val LabeledPoint(label, features) = iter.next
+ val yDiff = label - yMean
+ if (yDiff != 0) {
+features.iterator.zip(xMeans.iterator)
+ .foreach { case ((col, x), (_, xMean)) => array(col) += yDiff *
(x - xMean) }
+ }
+}
+Iterator.single(array)
+ } else Iterator.empty
+}.treeReduce { case (array1, array2) =>
+ var i = 0
+ while (i < numFeatures) {
+array1(i) += array2(i)
+i += 1
+ }
+ array1
+}
+
+for(i <- 0 until numFeatures) {
+ // Cov(X,Y) = Sum(((Xi - Avg(X)) * ((Yi-Avg(Y))) / (N-1)
+ val covariance = sumForCov (i) / (numSamples - 1)
+ val corr = covariance / (yStd * xStd(i))
+ val fValue = corr * corr / (1 - corr * corr) * degreeOfFreedom
+ val pValue = 1.0 - new FDistribution(1,
degreeOfFreedom).cumulativeProbability(fValue)
Review comment:
What about reusing `new FDistribution(1, degreeOfFreedom)` as a variable:
```
val fd = new FDistribution(1, degreeOfFreedom)
for(...){
...
val pValue = 1.0 - fd.cumulativeProbability(fValue)
...
}
```
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org
With regards,
Apache Git Services
[GitHub] [spark] zhengruifeng commented on a change in pull request #27623: [SPARK-30867][ML] Add FValueRegressionTest
zhengruifeng commented on a change in pull request #27623: [SPARK-30867][ML]
Add FValueRegressionTest
URL: https://github.com/apache/spark/pull/27623#discussion_r381066633
##
File path: mllib/src/main/scala/org/apache/spark/ml/stat/SelectionTest.scala
##
@@ -0,0 +1,100 @@
+/*
+ * 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.apache.commons.math3.distribution.FDistribution
+
+import org.apache.spark.annotation.Since
+import org.apache.spark.ml.feature.LabeledPoint
+import org.apache.spark.ml.linalg.{Vector, VectorUDT}
+import org.apache.spark.ml.util.SchemaUtils
+import org.apache.spark.sql.{Dataset, Row}
+import org.apache.spark.sql.functions.{avg, col, stddev}
+
+@Since("3.1.0")
+object SelectionTest {
+
+ /**
+ * @param dataset DataFrame of continuous labels and continuous features.
+ * @param featuresCol Name of features column in dataset, of type `Vector`
(`VectorUDT`)
+ * @param labelCol Name of label column in dataset, of any numerical type
+ * @return Array containing the SelectionTestResult for every feature
against the label.
+ */
+ @Since("3.1.0")
+ def fValueRegressionTest(dataset: Dataset[_], featuresCol: String, labelCol:
String):
+Array[SelectionTestResult] = {
+
+val spark = dataset.sparkSession
+import spark.implicits._
+
+SchemaUtils.checkColumnType(dataset.schema, featuresCol, new VectorUDT)
+SchemaUtils.checkNumericType(dataset.schema, labelCol)
+
+val Row(xMeans: Vector, xStd: Vector, yMean: Double, yStd: Double, count:
Long) = dataset
+ .select(Summarizer.metrics("mean", "std",
"count").summary(col(featuresCol)).as("summary"),
+avg(col(labelCol)).as("yMean"),
+stddev(col(labelCol)).as("yStd"))
+ .select("summary.mean", "summary.std", "yMean", "yStd", "summary.count")
+ .first()
+
+val labeledPointRdd = dataset.select(col("label").cast("double"),
col("features"))
+ .as[(Double, Vector)]
+ .rdd.map { case (label, features) => LabeledPoint(label, features) }
Review comment:
it seems we do not need to convert `(label, features)` to `LabeledPoint`,
since this algorithm is not implemented in the `.mllib` side like other
`ChiSquareTest`
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org
With regards,
Apache Git Services
-
To unsubscribe, e-mail: reviews-unsubscr...@spark.apache.org
For additional commands, e-mail: reviews-h...@spark.apache.org
