Github user yanboliang commented on a diff in the pull request:
https://github.com/apache/spark/pull/15851#discussion_r87693157
--- Diff: R/pkg/inst/tests/testthat/test_mllib.R ---
@@ -971,10 +971,15 @@ test_that("spark.randomForest Classification", {
predictions <- collect(predict(model, data))$prediction
expect_equal(length(grep("1.0", predictions)), 50)
expect_equal(length(grep("2.0", predictions)), 50)
+
+ # spark.randomForest classification can work on libsvm data
+ data <-
read.df(absoluteSparkPath("data/mllib/sample_multiclass_classification_data.txt"),
+ source = "libsvm")
+ model <- spark.randomForest(data, label ~ features, "classification")
+ expect_equal(summary(model)$numFeatures, 4)
--- End diff --
The reason behind this is there is no suitable libsvm data under
```data/mllib/``` for ```naiveBayes``` and ```glm```. Currently SparkR only
support ```Bernoulli``` naive bayes which require all features data should be 0
or 1. And the binary classification data under ```data/mllib``` is collinear
data, which is not very suitable to illustrate ```glm``` with binomial family.
I'm working on another PR to fix this collinear issue, so I added test for
```randomForest``` and ```gbt```.
---
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 [email protected] or file a JIRA ticket
with INFRA.
---
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
