[jvm-packages] enable predictLeaf/predictContrib/treeLimit in 0.8

[CodingCat]

missing the test cases

[codecov-io]

Codecov Report

Merging #3532 into master will increase coverage by 0.21%.
The diff coverage is 76.41%.

@@             Coverage Diff              @@
##             master    #3532      +/-   ##
============================================
+ Coverage     45.64%   45.86%   +0.21%     
  Complexity      188      188              
============================================
  Files           166      166              
  Lines         13234    13320      +86     
  Branches        445      468      +23     
============================================
+ Hits           6041     6109      +68     
- Misses         6989     6994       +5     
- Partials        204      217      +13

Impacted Files	Coverage Δ	Complexity Δ
...c/main/scala/ml/dmlc/xgboost4j/scala/Booster.scala	35.71% <50%> (ø)	8 <1> (ø)	⬇️
...c/xgboost4j/scala/spark/params/GeneralParams.scala	66.66% <66.66%> (ø)	0 <0> (ø)	⬇️
...c/xgboost4j/scala/spark/params/BoosterParams.scala	58.82% <75%> (+0.28%)	0 <0> (ø)	⬇️
...dmlc/xgboost4j/scala/spark/XGBoostClassifier.scala	66.53% <76.47%> (+2.62%)	18 <0> (ø)	⬇️
.../dmlc/xgboost4j/scala/spark/XGBoostRegressor.scala	65.68% <79.06%> (+3.41%)	16 <0> (ø)	⬇️

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 6cf97b4...dc99d66. Read the comment docs.

[hcho3]

@CodingCat Should we shoot for merging this PR before releasing 0.80? If we need to, we can delay 0.80 release a bit.

[CodingCat]

@hcho3 , yeah this is a blocking issue for the next release,

too busy in daily jobs and also that tutorial thing, need more time (expectation is Friday) to add unit test cases here

[CodingCat]

@yanboliang would you please take a look?

[yanboliang]

Looks pretty good overall, I left one minor comments.

[yanboliang]

I have a little concern that users may set leafPredictionCol with "", their intention should be not to output this column, but here we don't handle this case. At MLlib, we check $(leafPredictionCol).isEmpty as well. What do you think about it?

[CodingCat]

yes, good point, let me revise it

[CodingCat]

@yanboliang please check the latest update

[yanboliang]

Looks good, it seams the failure is not relevant.

[hcho3]

I fixed the error in the Jenkins build. It should build successfully now.

[CodingCat]

@hcho3 I merged this and cherry-picked to release_0.80 branch

[hcho3]

@CodingCat Okay, I will do a code freeze using this commit. The branch will be re-based against the master at this commit.

As for the doc building, I'm going to place the build script on the Jenkins CI server. This way, we can install Java, Scala, and Maven. The built Javadoc will be placed to an S3 bucket, which will be then downloaded by ReadTheDocs. Expect a commit by end of today.

[joyang1]

Why Xgboost4j predictLeaf method is always have 3 leaves ?