Merge pull request #23 from IBM/pipe_or_expression

support :| operation for selecting best learners

Author: ppalmes

src/AutoMLPipeline.jl

@@ -14,11 +14,6 @@ include("basefilters.jl")
 using .BaseFilters
 export OneHotEncoder
 
-include("pipelines.jl")
-using .Pipelines
-export @pipeline
-export @pipelinex
-export Pipeline, ComboPipeline 
 
 include("featureselector.jl")
 using .FeatureSelectors
@@ -48,4 +43,10 @@ include("skcrossvalidator.jl")
 using .SKCrossValidators
 export crossvalidate
 
+include("pipelines.jl")
+using .Pipelines
+export @pipeline
+export @pipelinex
+export Pipeline, ComboPipeline 
+
 end # module

src/featureselector.jl

@@ -42,11 +42,11 @@ mutable struct FeatureSelector <: Transformer
     end
 end
 
-```
+"""
     FeatureSelector(cols::Vector{Int})
 
 Helper function for FeatureSelector.
-```
+"""
 function FeatureSelector(cols::Vector{Int})
     FeatureSelector(Dict(:columns => cols))
 end
@@ -195,11 +195,11 @@ mutable struct CatNumDiscriminator <: Transformer
     end
 end
 
-```
+"""
     CatNumDiscriminator(maxcat::Int)
 
 Helper function for CatNumDiscriminator.
-```
+"""
 function CatNumDiscriminator(maxcat::Int)
     CatNumDiscriminator(Dict(:maxcategories=>maxcat))
 end

src/pipelines.jl

@@ -6,6 +6,7 @@ using Random
 using AutoMLPipeline.AbsTypes
 using AutoMLPipeline.BaseFilters
 using AutoMLPipeline.Utils
+using AutoMLPipeline.EnsembleMethods: BestLearner
 
 import AutoMLPipeline.AbsTypes: fit!, transform!
 export fit!, transform!
@@ -189,6 +190,8 @@ function processexpr(args)
       args[ndx] = :Pipeline
     elseif args[ndx] == :+
       args[ndx] = :ComboPipeline
+    elseif args[ndx] == :|
+      args[ndx] = :VoteEnsemble
     end
   end
   return args

test/test_pipeline.jl

@@ -52,8 +52,12 @@ function test_pipeline()
   catf = CatFeatureSelector()
   numf = NumFeatureSelector()
   rf = RandomForest()
+  ada = Adaboost()
+  pt = PrunedTree()
   pcombo3 = @pipeline disc |> ((catf + numf) + (numf |> pca) + (numf |> ica) + (catf|>ohe)) |> rf
   (fit_transform!(pcombo3,X,Y)  .== Y) |> sum == 150
+  pcombo4 = @pipeline (numf |> pca) + (numf |> ica) |> (ada | rf | pt)
+  @test crossvalidate(pcombo4,X,Y).mean >= 0.90
 end
 @testset "Pipelines" begin
   Random.seed!(123)