[SPARK-23690][ML] Add handleinvalid to VectorAssembler#20829

Closed

yogeshg wants to merge 25 commits intoapache:masterfrom yogeshg:rformula_handleinvalid

+284-47

Commits on Mar 14, 2018

Better error for streaming dataframes, ensure non-null Vectors in first.
MrBago
committed
add NaN for null column
Yogesh Garg
committed
get lengths with a map
Yogesh Garg
committed
wip
Yogesh Garg
committed
wip
Yogesh Garg
committed
wip
Yogesh Garg
committed
Merge branch 'rformula_handleinvalid' into vectorAssemblerStuff
MrBago
authored
Merge fixes.
MrBago
committed

Commits on Mar 15, 2018

Merge pull request #2 from MrBago/vectorAssemblerStuff
yogeshg
authored
fix issues with this implementation
Yogesh Garg
committed
fix bugs; add tests
Yogesh Garg
committed
clean
Yogesh Garg
committed
edge case of all null column to be skipped handled iff no elements found
Yogesh Garg
committed
update logic for missing metadata
Yogesh Garg
committed

Commits on Mar 16, 2018

rafactor logic to simplify code paths
Yogesh Garg
committed

Commits on Mar 17, 2018

update logic as discussed with Bago
Yogesh Garg
committed

Commits on Mar 19, 2018

fix test cases, remove some conditionals, move to static methods in companion object
Yogesh Garg
committed
use the correct statics, update since
Yogesh Garg
committed

Commits on Mar 20, 2018

collect instead of cache
Yogesh Garg
committed

Commits on Mar 22, 2018

update test logic; bugfix for na.drop; add more docs, error messages and style
Yogesh Garg
committed
No need to compare with anything; just call assemble()
Yogesh Garg
committed

Commits on Apr 2, 2018

handle NaNs too; update tests; style
Yogesh Garg
committed
hint in comments
Yogesh Garg
committed
remaining style changes in review
Yogesh Garg
committed
restore old example
Yogesh Garg
committed