Merge pull request #283 from PolyMathOrg/add_shuffling

Add shuffling
PolyMathOrg · Dec 14, 2023 · a37feeb · a37feeb
2 parents f085adf + c4df4c1
commit a37feeb
Show file tree

Hide file tree

Showing 17 changed files with 542 additions and 404 deletions.
diff --git a/src/DataFrame-Tests/DataFrameAggrGroupTest.class.st b/src/DataFrame-Tests/DataFrameAggrGroupTest.class.st
@@ -4,7 +4,7 @@ Class {
 	#instVars : [
 		'df'
 	],
-	#category : #'DataFrame-Tests-Core'
+	#category : 'DataFrame-Tests-Core'
 }
 
 { #category : #running }

diff --git a/src/DataFrame-Tests/DataFrameHeadTailTest.class.st b/src/DataFrame-Tests/DataFrameHeadTailTest.class.st
@@ -5,7 +5,7 @@ Class {
 		'df',
 		'series'
 	],
-	#category : #'DataFrame-Tests-Core'
+	#category : 'DataFrame-Tests-Core'
 }
 
 { #category : #running }

diff --git a/src/DataFrame-Tests/DataFrameInternalTest.class.st b/src/DataFrame-Tests/DataFrameInternalTest.class.st
@@ -4,7 +4,7 @@ Class {
 	#instVars : [
 		'df'
 	],
-	#category : #'DataFrame-Tests-Core'
+	#category : 'DataFrame-Tests-Core'
 }
 
 { #category : #running }

diff --git a/src/DataFrame-Tests/DataFrameStatsTest.class.st b/src/DataFrame-Tests/DataFrameStatsTest.class.st
@@ -4,7 +4,7 @@ Class {
 	#instVars : [
 		'df'
 	],
-	#category : #'DataFrame-Tests-Core'
+	#category : 'DataFrame-Tests-Core'
 }
 
 { #category : #running }

diff --git a/src/DataFrame-Tests/DataFrameTest.class.st b/src/DataFrame-Tests/DataFrameTest.class.st
@@ -7,6 +7,62 @@ Class {
 	#category : #'DataFrame-Tests-Core'
 }
 
+{ #category : #private }
+DataFrameTest >> expectedShuffledDataFrameWithSeedOne [
+	"In theory, shuffling an array with a fixed random seed should produce the same result across different versions of Pharo and the same implementation of the random number generator. The purpose of using a fixed seed is to make the random number generation deterministic, meaning that given the same seed, the sequence of random numbers generated will be the same every time.
+	There is a difference in the RNG used in Pharo 12, which uses the primitive 231, and the previous Pharo versions, which used a native implementation in privateNextSeed.
+	The change was introduced in this commit: https://github.com/pharo-project/pharo/commit/bf22496dbd0996ee470c9f85a7cc076e01dff57f
+	So we answer different data frames as the result ordering changes because of this implementation change"
+
+	| expected |
+	expected := SystemVersion current major >= 12
+		ifTrue: [ 
+			(DataFrame withRows: #(
+				('Barcelona' 1.609 true)
+		   		('London' 8.788 false)
+				('Dubai' 2.789 true))
+					rowNames: #( 'A' 'C' 'B'))
+					yourself  ]
+		ifFalse: [ 
+			(DataFrame withRows: #(
+				('Dubai' 2.789 true)
+		   		('London' 8.788 false)
+				('Barcelona' 1.609 true))
+					rowNames: #('B' 'C' 'A'))
+					yourself ].
+	expected columnNames: #( 'City' 'Population' 'BeenThere' ).
+	^ expected 
+
+]
+
+{ #category : #private }
+DataFrameTest >> expectedShuffledDataFrameWithSeedTwo [
+	"In theory, shuffling an array with a fixed random seed should produce the same result across different versions of Pharo and the same implementation of the random number generator. The purpose of using a fixed seed is to make the random number generation deterministic, meaning that given the same seed, the sequence of random numbers generated will be the same every time.
+	There is a difference in the RNG used in Pharo 12, which uses the primitive 231, and the previous Pharo versions, which used a native implementation in privateNextSeed.
+	The change was introduced in this commit: https://github.com/pharo-project/pharo/commit/bf22496dbd0996ee470c9f85a7cc076e01dff57f
+	So we answer different data frames as the result ordering changes because of this implementation change"
+
+	| expected |
+	expected := SystemVersion current major >= 12
+		ifTrue: [ 
+			(DataFrame withRows: #(
+				('Dubai' 2.789 true)	
+				('Barcelona' 1.609 true)
+  		 		('London' 8.788 false) )
+					rowNames: #('B' 'A' 'C'))
+					yourself  ]
+		ifFalse: [ 
+			(DataFrame withRows: #(
+		   		('London' 8.788 false)			
+				('Barcelona' 1.609 true)
+				('Dubai' 2.789 true))
+					rowNames: #('C' 'A' 'B'))
+					yourself ].
+	expected columnNames: #( 'City' 'Population' 'BeenThere' ).
+	^ expected 
+
+]
+
 { #category : #running }
 DataFrameTest >> setUp [
 
@@ -920,6 +976,19 @@ DataFrameTest >> testColumns [
 	self assert: df columns equals: expectedCollection
 ]
 
+{ #category : #tests }
+DataFrameTest >> testColumnsAllBut [
+
+	| expectedDataFrame |
+	expectedDataFrame := DataFrame withRows: #( #( 'Barcelona' 1.609 ) #( 'Dubai' 2.789 ) #( 'London' 8.788 ) ).
+	expectedDataFrame rowNames: #( 'A' 'B' 'C' ).
+	expectedDataFrame columnNames: #( 'City' 'Population').
+
+	self 
+		assert: (df columnsAllBut: #(BeenThere)) 
+		equals: expectedDataFrame
+]
+
 { #category : #tests }
 DataFrameTest >> testColumnsAt [
 
@@ -5164,6 +5233,20 @@ DataFrameTest >> testSelectEmptyDataFrame [
 	self assert: actual equals: expected
 ]
 
+{ #category : #tests }
+DataFrameTest >> testShuffledWithSeed [
+
+	| expected |
+
+	expected := self expectedShuffledDataFrameWithSeedOne.
+	self assert: (df shuffleWithSeed: 1) equals: expected.
+
+	expected := self expectedShuffledDataFrameWithSeedTwo.
+	self assert: (df shuffleWithSeed: 2) equals: expected.
+
+
+]
+
 { #category : #tests }
 DataFrameTest >> testSortBy [
 

diff --git a/src/DataFrame-Tests/DataPearsonCorrelationMethodTest.class.st b/src/DataFrame-Tests/DataPearsonCorrelationMethodTest.class.st
@@ -1,7 +1,7 @@
 Class {
 	#name : #DataPearsonCorrelationMethodTest,
 	#superclass : #TestCase,
-	#category : #'DataFrame-Tests-Math'
+	#category : 'DataFrame-Tests-Math'
 }
 
 { #category : #tests }

diff --git a/src/DataFrame-Tests/DataSeriesTest.class.st b/src/DataFrame-Tests/DataSeriesTest.class.st
@@ -5,7 +5,7 @@ Class {
 		'series',
 		'keyArray'
 	],
-	#category : #'DataFrame-Tests-Core'
+	#category : 'DataFrame-Tests-Core'
 }
 
 { #category : #running }

diff --git a/src/DataFrame/Array.extension.st b/src/DataFrame/Array.extension.st
@@ -1,6 +1,6 @@
-Extension { #name : #Array }
+Extension { #name : 'Array' }
 
-{ #category : #'*DataFrame' }
+{ #category : '*DataFrame' }
 Array >> calculateDataType [
 
 	| types |
@@ -18,7 +18,7 @@ Array >> calculateDataType [
 	^ UndefinedObject
 ]
 
-{ #category : #'*DataFrame' }
+{ #category : '*DataFrame' }
 Array >> leastCommonSuperclassOf: firstClass and: secondClass [
 "Determines the closest element of class hierarchy which is the common ancestor of two given classes"
 
@@ -40,7 +40,7 @@ Array >> leastCommonSuperclassOf: firstClass and: secondClass [
 	^ Object
 ]
 
-{ #category : #'*DataFrame' }
+{ #category : '*DataFrame' }
 Array >> sortIfPossible [
 	"Sort if possible"
 

diff --git a/src/DataFrame/Behavior.extension.st b/src/DataFrame/Behavior.extension.st
@@ -1,6 +1,6 @@
-Extension { #name : #Behavior }
+Extension { #name : 'Behavior' }
 
-{ #category : #'*DataFrame-Core-Base' }
+{ #category : '*DataFrame-Core-Base' }
 Behavior >> inheritsFromOrEqualTo: aClass [
 	"Answer whether the argument, aClass, is equal to the receiver or belongs to its superclass chain."
 

diff --git a/src/DataFrame/Collection.extension.st b/src/DataFrame/Collection.extension.st
@@ -1,12 +1,12 @@
-Extension { #name : #Collection }
+Extension { #name : 'Collection' }
 
-{ #category : #'*DataFrame-Core-Base' }
+{ #category : '*DataFrame-Core-Base' }
 Collection >> ** arg [
 
 	^ self raisedTo: arg
 ]
 
-{ #category : #'*DataFrame' }
+{ #category : '*DataFrame' }
 Collection >> asDataFrame [
 
 	| numberOfRows numberOfColumns dataFrame |
@@ -31,26 +31,26 @@ Collection >> asDataFrame [
 	^ dataFrame
 ]
 
-{ #category : #'*DataFrame-Core-Base' }
+{ #category : '*DataFrame-Core-Base' }
 Collection >> asDataSeries [
 
 	^ DataSeries newFrom: self
 ]
 
-{ #category : #'*DataFrame-Core-Base' }
+{ #category : '*DataFrame-Core-Base' }
 Collection >> closeTo: aCollection [
 
 	^ (self - aCollection) inject: true into: [ :accum :each |
 		accum and: (each closeTo: 0) ]
 ]
 
-{ #category : #'*DataFrame-Core-Base' }
+{ #category : '*DataFrame-Core-Base' }
 Collection >> variance [
 
 	^ self stdev squared
 ]
 
-{ #category : #'*DataFrame' }
+{ #category : '*DataFrame' }
 Collection >> withSeries: aDataSeries collect: twoArgBlock [
 	"Collect and return the result of evaluating twoArgBlock with corresponding elements from this collection and aDataSeries."
 	| result |

diff --git a/src/DataFrame/DataCorrelationMethod.class.st b/src/DataFrame/DataCorrelationMethod.class.st
@@ -1,10 +1,12 @@
 Class {
-	#name : #DataCorrelationMethod,
-	#superclass : #Object,
-	#category : #'DataFrame-Math'
+	#name : 'DataCorrelationMethod',
+	#superclass : 'Object',
+	#category : 'DataFrame-Math',
+	#package : 'DataFrame',
+	#tag : 'Math'
 }
 
-{ #category : #comparing }
+{ #category : 'comparing' }
 DataCorrelationMethod class >> between: x and: y [
 	"Calcualte the correlation coefficient between two data series"
 	self subclassResponsibility