8b08e14de7
## What changes were proposed in this pull request? some cleanup and tiny optimization 1, since the `transformImpl` method in the .mllib side is no longer used in the .ml side, the scope should be limited; 2, in the `hashUDF`, val `numOfFeatures` is never used; 3, in the udf, it is inefficient to involve param getter (`$(numFeatures)`/`$(binary)`) directly or via method `indexOf` ((`$(numFeatures)`) . instead, the getter should be called outside of the udf; ## How was this patch tested? existing suites Closes #25324 from zhengruifeng/hashingtf_cleanup. Authored-by: zhengruifeng <ruifengz@foxmail.com> Signed-off-by: Sean Owen <sean.owen@databricks.com> |
||
---|---|---|
.. | ||
benchmarks | ||
src | ||
pom.xml |