add when column method

[frosforever]

connects to #164

Differs from the spark implementation in that it requires an otherwise and will not default to null.

[codecov-io]

Codecov Report

Merging #220 into master will decrease coverage by 0.22%.
The diff coverage is 100%.

@@            Coverage Diff             @@
##           master     #220      +/-   ##
==========================================
- Coverage    96.1%   95.87%   -0.23%     
==========================================
  Files          52       42      -10     
  Lines         924      849      -75     
  Branches       12       13       +1     
==========================================
- Hits          888      814      -74     
+ Misses         36       35       -1

Impacted Files	Coverage Δ
...la/frameless/functions/NonAggregateFunctions.scala	100% <100%> (ø)	⬆️
...c/main/scala/frameless/TypedDatasetForwarded.scala	50% <0%> (-4.17%)	⬇️
...ataset/src/main/scala/frameless/TypedDataset.scala	93.7% <0%> (-0.2%)	⬇️
dataset/src/main/scala/frameless/TypedColumn.scala	100% <0%> (ø)	⬆️
...src/main/scala/frameless/ml/TypedTransformer.scala
...la/frameless/ml/feature/TypedVectorAssembler.scala
...cala/frameless/ml/feature/TypedStringIndexer.scala
...la/frameless/ml/internals/UnaryInputsChecker.scala
...scala/frameless/ml/internals/SelectorByValue.scala
... and 5 more

---

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 f249172...60f0df0. Read the comment docs.

[imarios]

LGTM!

[imarios]

can we add an example here?

[frosforever]

@imarios yet again thanks for your review! Comments addressed.

[imarios]

Hey @frosforever

So I played a bit with the code and came up with what I believe is a simplified implementation.

   ... rest of NonAggregateFunctions
   /** Non-Aggregate function: Evaluates a list of conditions and returns one of multiple
    * possible result expressions. If none match, otherwise is returned
    * {{{
    *   when(ds('boolField), ds('a))
    *     .when(ds('otherBoolField), lit(123))
    *     .otherwise(ds('b))
    * }}}
    * apache/spark
    */
  def when[T, A](condition: TypedColumn[T, Boolean], value: TypedColumn[T, A]): When[T, A] =
    new When[T, A](untyped.when(condition.untyped, value.untyped))
}

private[functions] class When[T, A](untypedC: Column) {
  def when(condition: TypedColumn[T, Boolean], value: TypedColumn[T, A]): When[T, A] = new When[T, A](
    untypedC.when(condition.untyped, value.untyped)
  )

  def otherwise(value: TypedColumn[T, A]): TypedColumn[T, A] =
    new TypedColumn[T, A](untypedC.otherwise(value.untyped))(value.uencoder)
}

What do you think?

[frosforever]

@imarios Ha! that's nearly identical to a version I had earlier! Swapped it out for the one I pushed up for some reason that I can't seem to recall anymore...
Mine had an additional constructor instead

[OlivierBlanvillain]

The #218 merge created some conflicts 😕

[imarios]

@frosforever can you resolve conflicts rebase/squash? I will merge as soon as everything clears from CI.

[frosforever]

Thanks for the reviews @imarios and @OlivierBlanvillain! Conflicts resolved

[OlivierBlanvillain]

@frosforever Thanks for the PR! 👍