zhengruifeng commented on a change in pull request #27982: [SPARK-31222][ML]
Make ANOVATest Sparsity-Aware
URL: https://github.com/apache/spark/pull/27982#discussion_r396222712
##########
File path: mllib/src/test/scala/org/apache/spark/ml/stat/ANOVATestSuite.scala
##########
@@ -144,22 +144,30 @@ class ANOVATestSuite
}
test("test DataFrame with sparse vector") {
- val df = spark.createDataFrame(Seq(
- (3, Vectors.sparse(6, Array((0, 6.0), (1, 7.0), (3, 7.0), (4, 6.0)))),
- (1, Vectors.sparse(6, Array((1, 9.0), (2, 6.0), (4, 5.0), (5, 9.0)))),
- (3, Vectors.sparse(6, Array((1, 9.0), (2, 3.0), (4, 5.0), (5, 5.0)))),
- (2, Vectors.dense(Array(0.0, 9.0, 8.0, 5.0, 6.0, 4.0))),
- (2, Vectors.dense(Array(8.0, 9.0, 6.0, 5.0, 4.0, 4.0))),
- (3, Vectors.dense(Array(8.0, 9.0, 6.0, 4.0, 0.0, 0.0)))
- )).toDF("label", "features")
+ val data = Seq(
+ (3, Vectors.dense(Array(6.0, 7.0, 0.0, 7.0, 6.0, 0.0, 0.0))),
+ (1, Vectors.dense(Array(0.0, 9.0, 6.0, 0.0, 5.0, 9.0, 0.0))),
+ (3, Vectors.dense(Array(0.0, 9.0, 3.0, 0.0, 5.0, 5.0, 0.0))),
+ (2, Vectors.dense(Array(0.0, 9.0, 8.0, 5.0, 6.0, 4.0, 0.0))),
+ (2, Vectors.dense(Array(8.0, 9.0, 6.0, 5.0, 4.0, 4.0, 0.0))),
+ (3, Vectors.dense(Array(8.0, 9.0, 6.0, 4.0, 0.0, 0.0, 0.0))))
- val anovaResult = ANOVATest.test(df, "features", "label")
- val (pValues: Vector, fValues: Vector) =
- anovaResult.select("pValues", "fValues")
- .as[(Vector, Vector)].head()
- assert(pValues ~== Vectors.dense(0.71554175, 0.71554175, 0.34278574,
0.45824059, 0.84633632,
- 0.15673368) relTol 1e-6)
- assert(fValues ~== Vectors.dense(0.375, 0.375, 1.5625, 1.02364865,
0.17647059,
- 3.66) relTol 1e-6)
+ val df1 = spark.createDataFrame(data.map(t => (t._1, t._2.toDense)))
+ .toDF("label", "features")
+ val df2 = spark.createDataFrame(data.map(t => (t._1, t._2.toSparse)))
+ .toDF("label", "features")
+ val df3 = spark.createDataFrame(data.map(t => (t._1, t._2.compressed)))
+ .toDF("label", "features")
+
+ Seq(df1, df2, df3).foreach { df =>
+ val anovaResult = ANOVATest.test(df, "features", "label")
+ val (pValues: Vector, fValues: Vector) =
+ anovaResult.select("pValues", "fValues")
+ .as[(Vector, Vector)].head()
+ assert(pValues ~== Vectors.dense(0.71554175, 0.71554175, 0.34278574,
0.45824059, 0.84633632,
+ 0.15673368, Double.NaN) relTol 1e-6)
Review comment:
for column only containing zero values, sklearn also returns `nan`:
```python
X = np.zeros([3,5])
y = [1,2,3]
f_classif(X, y)
/home/zrf/Applications/anaconda3/lib/python3.7/site-packages/sklearn/feature_selection/_univariate_selection.py:110:
RuntimeWarning: invalid value encountered in true_divide
msw = sswn / float(dfwn)
Out[24]: (array([nan, nan, nan, nan, nan]), array([nan, nan, nan, nan, nan]))
=
```
----------------------------------------------------------------
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:
[email protected]
With regards,
Apache Git Services
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
