Add shuffle data frame support

Add split train test method
pharo-ai · Dec 11, 2023 · d49e02c · d49e02c
1 parent 182681f
commit d49e02c
Show file tree

Hide file tree

Showing 6 changed files with 67 additions and 28 deletions.
diff --git a/src/AI-DataPartitioners/AIHashPartitioner.class.st b/src/AI-DataPartitioners/AIHashPartitioner.class.st
@@ -30,18 +30,19 @@ subsets := AIHashPartitioner split: dataFrame withProportions: #(0.8 0.2).
 ```
 "
 Class {
-	#name : #AIHashPartitioner,
-	#superclass : #Object,
-	#category : #'AI-DataPartitioners'
+	#name : 'AIHashPartitioner',
+	#superclass : 'Object',
+	#category : 'AI-DataPartitioners',
+	#package : 'AI-DataPartitioners'
 }
 
-{ #category : #api }
+{ #category : 'api' }
 AIHashPartitioner class >> split: aCollection withProportions: aCollectionOfProportions [
 
 	^ self new split: aCollection withProportions: aCollectionOfProportions
 ]
 
-{ #category : #api }
+{ #category : 'api' }
 AIHashPartitioner >> split: aCollection withProportions: aCollectionOfProportions [
 
 	| cumulativeProportionsMap indexesMap |

diff --git a/src/AI-DataPartitioners/AIRandomPartitioner.class.st b/src/AI-DataPartitioners/AIRandomPartitioner.class.st
@@ -15,50 +15,61 @@ subsets := AIRandomPartitioner split: letters withProportions: #(0.5 0.3 0.2).
 ```
 "
 Class {
-	#name : #AIRandomPartitioner,
-	#superclass : #Object,
-	#category : #'AI-DataPartitioners'
+	#name : 'AIRandomPartitioner',
+	#superclass : 'Object',
+	#category : 'AI-DataPartitioners',
+	#package : 'AI-DataPartitioners'
 }
 
-{ #category : #api }
+{ #category : 'api' }
 AIRandomPartitioner class >> split: aCollection withBoundaries: aCollectionOfBoundaries [
 
 	^ self new split: aCollection withBoundaries: aCollectionOfBoundaries
 ]
 
-{ #category : #api }
+{ #category : 'api' }
 AIRandomPartitioner class >> split: aCollection withBoundaries: aCollectionOfBoundaries seed: aNumber [
 
 	^ self new split: aCollection withBoundaries: aCollectionOfBoundaries seed: aNumber
 ]
 
-{ #category : #api }
+{ #category : 'api' }
 AIRandomPartitioner class >> split: aCollection withProportions: aCollectionOfProportions [
 
 	^ self new split: aCollection withProportions: aCollectionOfProportions
 ]
 
-{ #category : #api }
+{ #category : 'api' }
 AIRandomPartitioner class >> split: aCollection withProportions: aCollectionOfProportions seed: aNumber [
 
 	^ self new split: aCollection withProportions: aCollectionOfProportions seed: aNumber
 ]
 
-{ #category : #api }
+{ #category : 'api' }
+AIRandomPartitioner class >> split: aCollection withProportions: aCollectionOfProportions shuffle: aBoolean [
+	"Split aCollection into multiple sets. The sets to split are determined by the size of aCollectionOfProportions, which looks like #(0.7 0.3), and commonly it represents splitting between a training and test set."
+
+	^ self new 
+		split: aCollection 
+		withProportions: aCollectionOfProportions 
+		shuffle: aBoolean
+]
+
+{ #category : 'api' }
 AIRandomPartitioner class >> split: aCollection withSizes: aCollectionOfSizes [
 
 	^ self new split: aCollection withSizes: aCollectionOfSizes
 ]
 
-{ #category : #accessing }
+{ #category : 'accessing' }
 AIRandomPartitioner >> shuffle: aCollection [
 
 	"Default method of shuffling the elements of collection. You can override it if you need to use a different shuffling algorithm"
 
-	^ aCollection shuffle
+	^ aCollection shuffled
 ]
 
-{ #category : #accessing }
+{ #category : 'accessing' }
 AIRandomPartitioner >> shuffle: aCollection seed: aNumber [
 	"Default method of shuffling the elements of collection. You can override it if you need to use a different shuffling algorithm"
 
@@ -67,13 +78,13 @@ AIRandomPartitioner >> shuffle: aCollection seed: aNumber [
 	^ aCollection shuffleWithSeed: aNumber
 ]
 
-{ #category : #private }
+{ #category : 'private' }
 AIRandomPartitioner >> split: aCollection withBoundaries: aCollectionOfBoundaries [
 
 	^ self split: aCollection withBoundaries: aCollectionOfBoundaries seed: nil
 ]
 
-{ #category : #private }
+{ #category : 'private' }
 AIRandomPartitioner >> split: aCollection withBoundaries: aCollectionOfBoundaries seed: aNumber [
 
 	| shuffledIndices previousBoundary |
@@ -88,13 +99,13 @@ AIRandomPartitioner >> split: aCollection withBoundaries: aCollectionOfBoundarie
 		  res ]
 ]
 
-{ #category : #api }
+{ #category : 'api' }
 AIRandomPartitioner >> split: aCollection withProportions: aCollectionOfProportions [
 
 	^ self split: aCollection withProportions: aCollectionOfProportions seed: nil
 ]
 
-{ #category : #api }
+{ #category : 'api' }
 AIRandomPartitioner >> split: aCollection withProportions: aCollectionOfProportions seed: aNumber [
 	"If you want to use a fixed seed for the random numbers generator for have reproducibility"
 
@@ -107,7 +118,19 @@ AIRandomPartitioner >> split: aCollection withProportions: aCollectionOfProporti
 	^ self split: aCollection withBoundaries: boundaries seed: aNumber
 ]
 
-{ #category : #api }
+{ #category : 'api' }
+AIRandomPartitioner >> split: aCollection withProportions: aCollectionOfProportions shuffle: aBoolean [
+	"See class side comment."
+
+	| targetCollection |
+
+	targetCollection := aBoolean 
+		ifTrue: [ self shuffle: aCollection ]
+		ifFalse: [ aCollection ].
+	^ self split: targetCollection withProportions: aCollectionOfProportions seed: nil
+]
+
+{ #category : 'api' }
 AIRandomPartitioner >> split: aCollection withSizes: aCollectionOfSizes [
 
 	| boundaries |

diff --git a/src/AI-DataPartitioners/Collection.extension.st b/src/AI-DataPartitioners/Collection.extension.st
@@ -1,6 +1,6 @@
-Extension { #name : #Collection }
+Extension { #name : 'Collection' }
 
-{ #category : #'*AI-DataPartitioners' }
+{ #category : '*AI-DataPartitioners' }
 Collection >> cumulativeSum [
 	| sum |
 	sum := 0.

diff --git a/src/AI-DataPartitioners/DataFrame.extension.st b/src/AI-DataPartitioners/DataFrame.extension.st
@@ -0,0 +1,15 @@
+Extension { #name : 'DataFrame' }
+
+{ #category : '*AI-DataPartitioners' }
+DataFrame >> splitTrainTest: aCollectionOfProportions shuffle: aBoolean [ 
+	"Split the receiver's data into train and test sets.
+	
+	xTrain and yTrain sets are used for training and fitting the model.
+	xTest and yTest sets are used for testing the model.
+	"
+
+	^ AIRandomPartitioner 
+		split: self 
+		withProportions: aCollectionOfProportions 
+		shuffle: aBoolean.
+]
diff --git a/src/AI-DataPartitioners/SequenceableCollection.extension.st b/src/AI-DataPartitioners/SequenceableCollection.extension.st
@@ -1,18 +1,18 @@
-Extension { #name : #SequenceableCollection }
+Extension { #name : 'SequenceableCollection' }
 
-{ #category : #'*AI-DataPartitioners' }
+{ #category : '*AI-DataPartitioners' }
 SequenceableCollection >> indexOfRowNamed: aName [
 
 	^ self indexOf: aName
 ]
 
-{ #category : #'*AI-DataPartitioners' }
+{ #category : '*AI-DataPartitioners' }
 SequenceableCollection >> rowNames [
 
 	^ self
 ]
 
-{ #category : #'*AI-DataPartitioners' }
+{ #category : '*AI-DataPartitioners' }
 SequenceableCollection >> shuffleWithSeed: aNumber [
 	" Fix the random seed with aNumber to ensure reproducibility "
 

diff --git a/src/AI-DataPartitioners/package.st b/src/AI-DataPartitioners/package.st
@@ -1 +1 @@
-Package { #name : #'AI-DataPartitioners' }
+Package { #name : 'AI-DataPartitioners' }