[SPARK-13982][SparkR] KMean's predict: Features output column name is an unclear, automatically genetared text

[NarineK]

What changes were proposed in this pull request?

In KMean's predict the output column name of the features was set to something like this: "vecAssembler_522ba59ea239__output", which is the default output column name of the "VectorAssembler".
I set a value "features" which is more intuitive and human readable
Before fix:
predict(model, training)
DataFrame[Sepal_Length:double, Sepal_Width:double, Petal_Length:double, Petal_Width:double, vecAssembler_522ba59ea239__output: vector, prediction:int]

After fix:
DataFrame[Sepal_Length:double, Sepal_Width:double, Petal_Length:double, Petal_Width:double, features:vector, prediction:int]

How was this patch tested?

All previous test cases pass!

[SparkQA]

Test build #53464 has finished for PR 11793 at commit 48061de.

• This patch passes all tests.
• This patch merges cleanly.
• This patch adds no public classes.

[SparkQA]

Test build #54142 has finished for PR 11793 at commit 7900fcf.

• This patch fails R style tests.
• This patch merges cleanly.
• This patch adds no public classes.

[SparkQA]

Test build #54145 has finished for PR 11793 at commit 4101787.

• This patch passes all tests.
• This patch merges cleanly.
• This patch adds no public classes.

[NarineK]

@yanboliang, would you ,please, take a look at this ?

[yanboliang]

@NarineK This looks good to me. But I think we should wait the discussion of SPARK-14147, I vote that it should not output feature column in SparkR predict.

[jkbradley]

It looks like [https://github.com//pull/12039] solved this. Can you please verify and close this PR if so? Thanks!

[NarineK]

thanks, will close this.