1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
|
/*
* 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.regression
import org.apache.spark.SparkFunSuite
import org.apache.spark.ml.tree.impl.TreeTests
import org.apache.spark.ml.util.{DefaultReadWriteTest, MLTestingUtils}
import org.apache.spark.mllib.linalg.Vector
import org.apache.spark.mllib.regression.LabeledPoint
import org.apache.spark.mllib.tree.{DecisionTree => OldDecisionTree,
DecisionTreeSuite => OldDecisionTreeSuite}
import org.apache.spark.mllib.util.MLlibTestSparkContext
import org.apache.spark.rdd.RDD
import org.apache.spark.sql.{DataFrame, Row}
class DecisionTreeRegressorSuite
extends SparkFunSuite with MLlibTestSparkContext with DefaultReadWriteTest {
import DecisionTreeRegressorSuite.compareAPIs
private var categoricalDataPointsRDD: RDD[LabeledPoint] = _
override def beforeAll() {
super.beforeAll()
categoricalDataPointsRDD =
sc.parallelize(OldDecisionTreeSuite.generateCategoricalDataPoints())
}
/////////////////////////////////////////////////////////////////////////////
// Tests calling train()
/////////////////////////////////////////////////////////////////////////////
test("Regression stump with 3-ary (ordered) categorical features") {
val dt = new DecisionTreeRegressor()
.setImpurity("variance")
.setMaxDepth(2)
.setMaxBins(100)
.setSeed(1)
val categoricalFeatures = Map(0 -> 3, 1 -> 3)
compareAPIs(categoricalDataPointsRDD, dt, categoricalFeatures)
}
test("Regression stump with binary (ordered) categorical features") {
val dt = new DecisionTreeRegressor()
.setImpurity("variance")
.setMaxDepth(2)
.setMaxBins(100)
val categoricalFeatures = Map(0 -> 2, 1 -> 2)
compareAPIs(categoricalDataPointsRDD, dt, categoricalFeatures)
}
test("copied model must have the same parent") {
val categoricalFeatures = Map(0 -> 2, 1 -> 2)
val df = TreeTests.setMetadata(categoricalDataPointsRDD, categoricalFeatures, numClasses = 0)
val model = new DecisionTreeRegressor()
.setImpurity("variance")
.setMaxDepth(2)
.setMaxBins(8).fit(df)
MLTestingUtils.checkCopy(model)
}
test("predictVariance") {
val dt = new DecisionTreeRegressor()
.setImpurity("variance")
.setMaxDepth(2)
.setMaxBins(100)
.setPredictionCol("")
.setVarianceCol("variance")
val categoricalFeatures = Map(0 -> 2, 1 -> 2)
val df = TreeTests.setMetadata(categoricalDataPointsRDD, categoricalFeatures, numClasses = 0)
val model = dt.fit(df)
val predictions = model.transform(df)
.select(model.getFeaturesCol, model.getVarianceCol)
.collect()
predictions.foreach { case Row(features: Vector, variance: Double) =>
val expectedVariance = model.rootNode.predictImpl(features).impurityStats.calculate()
assert(variance === expectedVariance,
s"Expected variance $expectedVariance but got $variance.")
}
}
test("Feature importance with toy data") {
val dt = new DecisionTreeRegressor()
.setImpurity("variance")
.setMaxDepth(3)
.setSeed(123)
// In this data, feature 1 is very important.
val data: RDD[LabeledPoint] = TreeTests.featureImportanceData(sc)
val categoricalFeatures = Map.empty[Int, Int]
val df: DataFrame = TreeTests.setMetadata(data, categoricalFeatures, 0)
val model = dt.fit(df)
val importances = model.featureImportances
val mostImportantFeature = importances.argmax
assert(mostImportantFeature === 1)
assert(importances.toArray.sum === 1.0)
assert(importances.toArray.forall(_ >= 0.0))
}
test("should support all NumericType labels and not support other types") {
val dt = new DecisionTreeRegressor().setMaxDepth(1)
MLTestingUtils.checkNumericTypes[DecisionTreeRegressionModel, DecisionTreeRegressor](
dt, isClassification = false, sqlContext) { (expected, actual) =>
TreeTests.checkEqual(expected, actual)
}
}
/////////////////////////////////////////////////////////////////////////////
// Tests of model save/load
/////////////////////////////////////////////////////////////////////////////
test("read/write") {
def checkModelData(
model: DecisionTreeRegressionModel,
model2: DecisionTreeRegressionModel): Unit = {
TreeTests.checkEqual(model, model2)
assert(model.numFeatures === model2.numFeatures)
}
val dt = new DecisionTreeRegressor()
val rdd = TreeTests.getTreeReadWriteData(sc)
// Categorical splits with tree depth 2
val categoricalData: DataFrame =
TreeTests.setMetadata(rdd, Map(0 -> 2, 1 -> 3), numClasses = 0)
testEstimatorAndModelReadWrite(dt, categoricalData,
TreeTests.allParamSettings, checkModelData)
// Continuous splits with tree depth 2
val continuousData: DataFrame =
TreeTests.setMetadata(rdd, Map.empty[Int, Int], numClasses = 0)
testEstimatorAndModelReadWrite(dt, continuousData,
TreeTests.allParamSettings, checkModelData)
// Continuous splits with tree depth 0
testEstimatorAndModelReadWrite(dt, continuousData,
TreeTests.allParamSettings ++ Map("maxDepth" -> 0), checkModelData)
}
}
private[ml] object DecisionTreeRegressorSuite extends SparkFunSuite {
/**
* Train 2 decision trees on the given dataset, one using the old API and one using the new API.
* Convert the old tree to the new format, compare them, and fail if they are not exactly equal.
*/
def compareAPIs(
data: RDD[LabeledPoint],
dt: DecisionTreeRegressor,
categoricalFeatures: Map[Int, Int]): Unit = {
val numFeatures = data.first().features.size
val oldStrategy = dt.getOldStrategy(categoricalFeatures)
val oldTree = OldDecisionTree.train(data, oldStrategy)
val newData: DataFrame = TreeTests.setMetadata(data, categoricalFeatures, numClasses = 0)
val newTree = dt.fit(newData)
// Use parent from newTree since this is not checked anyways.
val oldTreeAsNew = DecisionTreeRegressionModel.fromOld(
oldTree, newTree.parent.asInstanceOf[DecisionTreeRegressor], categoricalFeatures)
TreeTests.checkEqual(oldTreeAsNew, newTree)
assert(newTree.numFeatures === numFeatures)
}
}
|
