Add a way to speedup dataframe data imputing

pharo-ai · Mar 21, 2023 · a43aa18 · a43aa18
1 parent 752871b
commit a43aa18
Show file tree

Hide file tree

Showing 2 changed files with 13 additions and 5 deletions.
diff --git a/src/AI-DataImputers/AISimpleImputer.class.st b/src/AI-DataImputers/AISimpleImputer.class.st
@@ -176,11 +176,7 @@ AISimpleImputer >> transform: aCollection [
 	self ensureIs2D: aCollection.
 	self statistics ifNil: [ self error: '#fit: needs to be called before transforming.' ].
 
-	^ aCollection collect: [ :subcoll |
-		  subcoll withIndexCollect: [ :elem :index |
-			  elem = self missingValue
-				  ifTrue: [ statistics at: index ]
-				  ifFalse: [ elem ] ] ]
+	^ aCollection copyReplace: missingValue in2DCollectionBy: statistics
 ]
 
 { #category : #options }

diff --git a/src/AI-DataImputers/SequenceableCollection.extension.st b/src/AI-DataImputers/SequenceableCollection.extension.st
@@ -0,0 +1,12 @@
+Extension { #name : #SequenceableCollection }
+
+{ #category : #'*AI-DataImputers' }
+SequenceableCollection >> copyReplace: missingValue in2DCollectionBy: arrayOfReplacementValues [
+	"I am a 2D collection and the goal is to return a copy replace the missing values by the values of my second parameter. The good value is the index of the missing value in the sub collection."
+
+	^ self collect: [ :subColl |
+		  subColl withIndexCollect: [ :element :index |
+			  element = missingValue
+				  ifTrue: [ arrayOfReplacementValues at: index ]
+				  ifFalse: [ element ] ] ]
+]