Merge pull request #94 from bensadeghi/cleanup

Little Cleanup

Author: bensadeghi

.travis.yml

@@ -5,7 +5,7 @@ os:
 
 julia:
   - 0.7
-  - 1.0
+  - 1.1
   - nightly
 
 matrix:

README.md

@@ -2,10 +2,7 @@
 
 [![Build Status](https://travis-ci.org/bensadeghi/DecisionTree.jl.svg?branch=master)](https://travis-ci.org/bensadeghi/DecisionTree.jl)
 [![Coverage Status](https://coveralls.io/repos/bensadeghi/DecisionTree.jl/badge.svg?branch=master)](https://coveralls.io/r/bensadeghi/DecisionTree.jl?branch=master)
-
-[![DecisionTree](http://pkg.julialang.org/badges/DecisionTree_0.5.svg)](http://pkg.julialang.org/?pkg=DecisionTree&ver=0.5)
-[![DecisionTree](http://pkg.julialang.org/badges/DecisionTree_0.6.svg)](http://pkg.julialang.org/?pkg=DecisionTree&ver=0.6)
-[![DecisionTree](http://pkg.julialang.org/badges/DecisionTree_0.7.svg)](http://pkg.julialang.org/?pkg=DecisionTree&ver=0.7)
+[![Docs Stable](https://img.shields.io/badge/docs-stable-blue.svg)](https://pkg.julialang.org/docs/DecisionTree/pEDeB/0.8.1/)
 
 Julia implementation of Decision Tree and Random Forest algorithms
 

src/classification/main.jl

@@ -342,71 +342,3 @@ function apply_adaboost_stumps_proba(stumps::Ensemble{S, T}, coeffs::Vector{Floa
                                     features::Matrix{S}, labels::Vector{T}) where {S, T}
     stack_function_results(row->apply_adaboost_stumps_proba(stumps, coeffs, row, labels), features)
 end
-
-
-#= temporarily commenting out new prune_tree implementation
-function prune_tree(tree::LeafOrNode{S, T}, purity_thresh=0.0) where {S, T}
-
-    function recursive_assign(leaf::Leaf{T}, set::Set{T})
-        for item in leaf.values
-            push!(set, item)
-        end
-    end
-
-    function recursive_assign(node::Node{S, T}, set::Set{T})
-        recursive_assign(node.left, set)
-        recursive_assign(node.right, set)
-    end
-
-    function recurse(
-            leaf          :: Leaf{T},
-            purity_thresh :: Float64,
-            label2int     :: Dict{T, Int},
-            labels        :: Vector{T})
-        nc = fill(0.0, length(labels))
-        for i in leaf.values
-            nc[label2int[i]] += 1.0
-        end
-        return nc, leaf
-    end
-
-    function recurse(
-            node          :: Node{S, T},
-            purity_thresh :: Float64,
-            label2int     :: Dict{T, Int},
-            labels        :: Vector{T})
-
-        ncl, l = recurse(node.left, purity_thresh, label2int, labels)
-        ncr, r = recurse(node.right, purity_thresh, label2int, labels)
-
-        if is_leaf(l) && is_leaf(r)
-
-            @simd for i in 1:length(labels)
-                ncl[i] += ncr[i]
-            end
-
-            n_samples = length(l.values) + length(r.values)
-            purity = -util.entropy(ncl, n_samples)
-            if purity > purity_thresh
-                return ncl, Leaf{T}(labels[argmax(ncl)], [l.values; r.values])
-            end
-
-        end
-
-        return ncl, Node{S, T}(node.featid, node.featval, l, r)
-    end
-
-    function main(tree::LeafOrNode{S, T}, purity_thresh=1.0)
-        set = Set{T}()
-        recursive_assign(tree, set)
-        labels = collect(set)
-        label2int  = Dict{T, Int}(label=>i for (i, label) in enumerate(labels))
-
-        ncl, node = recurse(tree, purity_thresh, label2int, labels)
-
-        return node
-    end
-
-    return main(tree, purity_thresh)
-end
-=#

src/regression/main.jl

@@ -93,43 +93,3 @@ function build_forest(
         return Ensemble{S, T}(forest)
     end
 end
-
-
-#= temporarily commenting out new prune_tree implementation
-function prune_tree(tree::LeafOrNode{S, T}, purity_thresh=0.0) where {S, T <: Float64}
-
-    function recurse(leaf :: Leaf{T}, purity_thresh :: Float64)
-        tssq = 0.0
-        tsum = 0.0
-        for v in leaf.values
-            tssq += v*v
-            tsum += v
-        end
-
-        return tssq, tsum, leaf
-    end
-
-    function recurse(node :: Node{S, T}, purity_thresh :: Float64)
-
-        lssq, lsum, l = recurse(node.left, purity_thresh)
-        rssq, rsum, r = recurse(node.right, purity_thresh)
-
-        if is_leaf(l) && is_leaf(r)
-            n_samples = length(l.values) + length(r.values)
-            tsum = lsum + rsum
-            tssq = lssq + rssq
-            tavg = tsum / n_samples
-            purity = tavg * tavg - tssq / n_samples
-            if purity > purity_thresh
-                return tsum, tssq, Leaf{T}(tavg, [l.values; r.values])
-            end
-
-        end
-
-        return 0.0, 0.0, Node{S, T}(node.featid, node.featval, l, r)
-    end
-
-    _, _, node = recurse(tree, purity_thresh)
-    return node
-end
-=#

test/local_runtests.jl

@@ -1,3 +1,56 @@
 using DecisionTree
+using DelimitedFiles
+using Distributed
+using Random
+using ScikitLearnBase
+using Statistics
+using Test
 
-include("test_suites.jl")
+println("Julia version: ", VERSION)
+
+function run_tests(list)
+    for test in list
+        println("TEST: $test \n")
+        include(test)
+        println("=" ^ 50)
+    end
+end
+
+classification = [
+    "classification/random.jl",
+    "classification/low_precision.jl",
+    "classification/heterogeneous.jl",
+    "classification/digits.jl",
+    "classification/iris.jl",
+    "classification/adult.jl",
+    "classification/scikitlearn.jl"
+]
+
+regression =     [
+    "regression/random.jl",
+    "regression/low_precision.jl",
+    "regression/digits.jl",
+    "regression/scikitlearn.jl"
+]
+
+miscellaneous =  [
+    "miscellaneous/parallel.jl"
+]
+
+test_suites = [
+    ("Classification", classification),
+    ("Regression", regression),
+    ("Miscellaneous", miscellaneous),
+]
+
+@testset "Test Suites" begin
+    for ts in 1:length(test_suites)
+        name = test_suites[ts][1]
+        list = test_suites[ts][2]
+        let
+            @testset "$name" begin
+                run_tests(list)
+            end
+        end
+    end
+end