Merge pull request #192 from jecisc/speed-up-creation-of-DataSeries

Speed up creation of data series

Author: jecisc

src/DataFrame-Tests/DataFrameTest.class.st

@@ -3725,25 +3725,18 @@ DataFrameTest >> testReplaceNilsWithMedian [
 { #category : #replacing }
 DataFrameTest >> testReplaceNilsWithMode [
 
-	|expected|
-
-	df := DataFrame withRows: #(
-		(7 1 nil)
-   		(8 nil nil)
-   		(nil 1 43)).
+	| expected |
+	df := DataFrame withRows: #( #( 7 1 nil ) #( 8 nil nil ) #( nil 1 43 ) ).
 
-	df rowNames: #(A B C).
-	df columnNames: #(Column1 Column2 Column3).
+	df rowNames: #( A B C ).
+	df columnNames: #( Column1 Column2 Column3 ).
 
-	expected := DataFrame withRows: #(
-		(7 1 43)
-   		(8 1 43)
-   		(7 1 43)).
+	expected := DataFrame withRows: #( #( 7 1 43 ) #( 8 1 43 ) #( 8 1 43 ) ).
 
-	expected rowNames: #(A B C).
-	expected columnNames: #(Column1 Column2 Column3).
+	expected rowNames: #( A B C ).
+	expected columnNames: #( Column1 Column2 Column3 ).
 
-	self assert: (df replaceNilsWithMode) equals: expected
+	self assert: df replaceNilsWithMode equals: expected
 ]
 
 { #category : #replacing }

src/DataFrame/DataFrame.class.st

@@ -1645,13 +1645,11 @@ DataFrame >> replaceNilsWithMedian [
 { #category : #replacing }
 DataFrame >> replaceNilsWithMode [
 
-	| modeOfColumn |
 	1 to: self numberOfColumns do: [ :i |
-		modeOfColumn := ((self columnAt: i) select: [ :ele | ele isNotNil ]) mode.
+		| modeOfColumn |
 		1 to: self numberOfRows do: [ :j |
-			(self at: j at: i) ifNil: [
-				self at: j at: i put: modeOfColumn  ] ].
-		]
+		(self at: j at: i) ifNil: [ self at: j at: i put: (modeOfColumn ifNil: [ modeOfColumn := ((self columnAt: i) select: [ :ele | ele isNotNil ]) mode ]) ] ].
+		modeOfColumn := nil ]
 ]
 
 { #category : #replacing }

src/DataFrame/DataSeries.class.st

@@ -21,6 +21,16 @@ DataSeries class >> newFrom: aCollection [
 		ifFalse: [ aCollection withIndexCollect: [ :each :i | i -> each ] ])
 ]
 
+{ #category : #'instance creation' }
+DataSeries class >> newFromKeys: keys andValues: values [
+
+	| dict |
+	self flag: #pharo12. "This is a copy of the superclass with a speed up. I'll propose this speedup in Pharo 12 so when Pharo 12 will be the minimal suuported version then we can drop this method."
+	dict := self new: keys size.
+	keys with: values do: [ :k :v | dict at: k put: v ].
+	^ dict
+]
+
 { #category : #'instance creation' }
 DataSeries class >> withKeys: keys values: values [
 	^ self newFromKeys: keys andValues: values
@@ -450,6 +460,7 @@ DataSeries >> makeNumerical [
 
 { #category : #accessing }
 DataSeries >> mode [
+
 	| valueCounts maxCount |
 	valueCounts := self valueCounts.
 	maxCount := valueCounts max.