zero323 commented on a change in pull request #27570: [SPARK-30820][SPARKR][ML]
Add FMClassifier to SparkR
URL: https://github.com/apache/spark/pull/27570#discussion_r387098974
##########
File path: R/pkg/tests/fulltests/test_mllib_classification.R
##########
@@ -488,4 +488,36 @@ test_that("spark.naiveBayes", {
expect_equal(class(collect(predictions)$clicked[1]), "character")
})
+test_that("spark.fmClassifier", {
+ df <- withColumn(
+ suppressWarnings(createDataFrame(iris)),
+ "Species", otherwise(when(column("Species") == "Setosa", "Setosa"),
"Not-Setosa")
+ )
+
+ model1 <- spark.fmClassifier(
+ df, Species ~ .,
+ regParam = 0.01, maxIter = 10, fitLinear = TRUE, factorSize = 3
+ )
+
+ prediction1 <- predict(model1, df)
+ expect_is(prediction1, "SparkDataFrame")
Review comment:
Sure we can. The question is what we are really trying to test in such
cases? What types of implementation mistakes can we detect here, that are not
already covered by JVM tests and / or SparkR data frames tests?
These checks involve additional jobs and many tests are already rejected to
keep things manageable, so unless these serve specific purpose, I'd prefer to
keep things lean here.
In contrast there are many SparkR ML failure modes that are real, and could
be tested, but are crippled by lack of required API. But that's way beyond the
scope of this PR.
----------------------------------------------------------------
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]
