59-DataSeries-should-be-a-subclass-of-OrderedDictionary

src/DataFrame-IO-Tests/DataFrameCsvReaderTest.class.st

@@ -6,7 +6,9 @@ Class {
 		'commaCsvFile',
 		'tabCsvFile',
 		'emptyCsvFile',
-		'expectedDataFrame'
+		'expectedDataFrame',
+		'dataFrameWithoutRowNames',
+		'dataFrameWithRowNames'
 	],
 	#category : #'DataFrame-IO-Tests'
 }
@@ -32,30 +34,55 @@ DataFrameCsvReaderTest >> setUp [
 	self createFile: tabCsvFile withContents: TestCsvStrings tabCsvString.
 	self createFile: emptyCsvFile withContents: TestCsvStrings emptyCsvString.
 
-	expectedDataFrame := DataFrame withRows: #(
+	dataFrameWithRowNames := DataFrame withRows: #(
 		(2.4 true 'rain')
 		(0.5 true 'rain')
 		(-1.2 true 'snow')
 		(-2.3 false '-')
 		(3.2 true 'rain')).
 
-	expectedDataFrame columnNames: #(temperature precipitation type).
-	expectedDataFrame rowNames: (#('01:10' '01:30' '01:50' '02:10' '02:30')
+	dataFrameWithRowNames columnNames: #(temperature precipitation type).
+	dataFrameWithRowNames rowNames: (#('01:10' '01:30' '01:50' '02:10' '02:30')
 		collect: #asTime).
+
+	dataFrameWithoutRowNames := DataFrame withRows: {
+		{ '01:10' asTime . 2.4 . true . 'rain' } .
+		{ '01:30' asTime . 0.5 . true . 'rain' } .
+		{ '01:50' asTime . -1.2 . true . 'snow' } .
+		{ '02:10' asTime . -2.3 . false . '-' } . 
+		{ '02:30' asTime . 3.2 . true . 'rain' }}.
+
+	dataFrameWithoutRowNames columnNames: #(nil temperature precipitation type).
 ]
 
 { #category : #tests }
 DataFrameCsvReaderTest >> testReadCsv [
+	| actualDataFrame |
+	actualDataFrame := DataFrame readFromCsv: commaCsvFile.
+	self assert: actualDataFrame equals: dataFrameWithoutRowNames.
+
+]
+
+{ #category : #tests }
+DataFrameCsvReaderTest >> testReadCsvWithRowNames [
 	| actualDataFrame |
 	actualDataFrame := DataFrame readFromCsvWithRowNames: commaCsvFile.
-	self assert: actualDataFrame equals: expectedDataFrame.
+	self assert: actualDataFrame equals: dataFrameWithRowNames.
 
 ]
 
 { #category : #tests }
-DataFrameCsvReaderTest >> testReadCsvWithSeparatorTab [
+DataFrameCsvReaderTest >> testReadCsvWithRowNamesWithSeparatorTab [
 	| actualDataFrame |
 	actualDataFrame := DataFrame readFromCsvWithRowNames: tabCsvFile separator: Character tab.
-	self assert: actualDataFrame equals: expectedDataFrame.
+	self assert: actualDataFrame equals: dataFrameWithRowNames.
+
+]
+
+{ #category : #tests }
+DataFrameCsvReaderTest >> testReadCsvWithSeparatorTab [
+	| actualDataFrame |
+	actualDataFrame := DataFrame readFromCsv: tabCsvFile withSeparator: Character tab.
+	self assert: actualDataFrame equals: dataFrameWithoutRowNames.
 
 ]

src/DataFrame-IO-Tests/DataFrameCsvWriterTest.class.st

@@ -48,6 +48,17 @@ DataFrameCsvWriterTest >> testWriteToCsv [
 	self assert: actual lines equals: expected lines.
 ]
 
+{ #category : #tests }
+DataFrameCsvWriterTest >> testWriteToCsvLineEndLf [
+	| writer actual expected |
+	writer := DataFrameCsvWriter new.
+	writer lineEndConvention: #lf.
+	dataFrame writeTo: commaQuoteCsvFile using: writer.
+	actual := self readFile: commaQuoteCsvFile.
+	expected := String lf join: TestCsvStrings commaQuoteCsvString lines.
+	self assert: actual lines equals: expected lines.
+]
+
 { #category : #tests }
 DataFrameCsvWriterTest >> testWriteToCsvWithSeparatorTab [
 	| actual expected |

src/DataFrame-IO/DataFrame.extension.st

@@ -10,23 +10,23 @@ DataFrame class >> readFrom: aLocation using: aDataFrameReader [
 DataFrame class >> readFromCsv: aFileReference [ 
 	| reader |
 	reader := DataFrameCsvReader new.
-	^ reader readFrom: aFileReference.
+	^ self readFrom: aFileReference using: reader.
 ]
 
 { #category : #'*DataFrame-IO' }
 DataFrame class >> readFromCsv: aFileReference withSeparator: aSeparator [ 
 	| reader |
 	reader := DataFrameCsvReader new.
 	reader separator: aSeparator.
-	^ reader readFrom: aFileReference.
+	^ self readFrom: aFileReference using: reader
 ]
 
 { #category : #'*DataFrame-IO' }
 DataFrame class >> readFromCsvWithRowNames: aFileReference [ 
 	| reader |
 	reader := DataFrameCsvReader new.
 	reader includeRowNames: true.
-	^ reader readFrom: aFileReference.
+	^ self readFrom: aFileReference using: reader
 ]
 
 { #category : #'*DataFrame-IO' }
@@ -35,7 +35,7 @@ DataFrame class >> readFromCsvWithRowNames: aFileReference separator: aSeparator
 	reader := DataFrameCsvReader new.
 	reader includeRowNames: true.
 	reader separator: aSeparator.
-	^ reader readFrom: aFileReference.
+	^ self readFrom: aFileReference using: reader
 ]
 
 { #category : #'*DataFrame-IO' }

src/DataFrame-Pharo6/OrderedDictionary.extension.st

@@ -0,0 +1,13 @@
+Extension { #name : #OrderedDictionary }
+
+{ #category : #'*DataFrame-Pharo6' }
+OrderedDictionary class >> newFromKeys: keys andValues: values [
+	"Create a dictionary from the keys and values arguments which should have the same length."
+
+	"(self newFromKeys: #(#x #y) andValues: #(3 6)) >>> (self new at: #x put: 3; at: #y put: 6 ;yourself)"
+
+	| dict |
+	dict := self new.
+	keys with: values do: [ :k :v | dict at: k put: v ].
+	^ dict
+]

src/DataFrame-Tests/DataFrameAggrGroupTest.class.st

@@ -25,24 +25,25 @@ DataFrameAggrGroupTest >> testAggregateAverage [
 
 	| expected actual |
 
-	expected := #(18.3433 20.79) asDataSeries.
-	expected keys: #(Male Female).
+	expected := DataSeries 
+		withKeys: #(Male Female)
+		values: #(18.3433 20.79)
+		name: #total_bill.
 
 	actual := (df group: #total_bill by: #sex) average.
-
 	self assert: actual closeTo: expected.
 ]
 
 { #category : #initialization }
 DataFrameAggrGroupTest >> testAggregateMax [
-
 	| expected actual |
 
-	expected := #(23.68 24.59) asDataSeries.
-	expected keys: #(Male Female).
+	expected := DataSeries 
+		withKeys: #(Male Female)
+		values: #(23.68 24.59)
+		name: #total_bill.
 
 	actual := (df group: #total_bill by: #sex) max.
-
 	self assert: actual equals: expected.
 ]
 
@@ -51,11 +52,12 @@ DataFrameAggrGroupTest >> testAggregateMin [
 
 	| expected actual |
 
-	expected := #(10.34 16.99) asDataSeries.
-	expected keys: #(Male Female).
+	expected := DataSeries 
+		withKeys: #(Male Female)
+		values: #(10.34 16.99)
+		name: #total_bill.
 
 	actual := (df group: #total_bill by: #sex) min.
-
 	self assert: actual equals: expected.
 ]
 
@@ -78,37 +80,19 @@ DataFrameAggrGroupTest >> testDataFrameGroupedPrintOn [
 	self assert: actual equals: expected.
 ]
 
-{ #category : #initialization }
-DataFrameAggrGroupTest >> testDataSeriesGroupedPrintOn [
-
-	| expected actual |
-
-	expected := String new writeStream.
-	expected
-		nextPutAll: 'a DataSeriesGrouped'; cr;
-		nextPutAll: ('Male: a DataSeries [3 items]'); cr;
-		nextPutAll: ('Female: a DataSeries [2 items]').
-	expected := expected contents.
-
-	actual := String new writeStream.
-	(df group: #total_bill by: #sex) printOn: actual.
-	actual := actual contents.
-
-	self assert: actual equals: expected.
-]
-
 { #category : #tests }
 DataFrameAggrGroupTest >> testGroupSeriesBySeries [
 
-	| femaleGroup maleGroup expectedSeries actualSeries |
+	| femaleGroup maleGroup expected actual |
 
 	femaleGroup := #(16.99 24.59) asDataSeries.
 	maleGroup := #(10.34 21.01 23.68) asDataSeries.
 
-	expectedSeries := { maleGroup . femaleGroup } asDataSeries.
-	expectedSeries keys: #(Male Female).
+	expected := DataSeries 
+		withKeys: #(Male Female)
+		values: { maleGroup . femaleGroup }
+		name: #total_bill.
 
-	actualSeries := (df group: #total_bill by: #sex) groups.
-
-self assert: actualSeries equals: expectedSeries.
+	actual := (df group: #total_bill by: #sex) groups.
+	self assert: actual equals: expected.
 ]

src/DataFrame-Tests/DataFrameHeadTailTest.class.st

@@ -104,64 +104,6 @@ DataFrameHeadTailTest >> testDataFrameTailN [
 	self assert: actual equals: expected.
 ]
 
-{ #category : #tests }
-DataFrameHeadTailTest >> testDataSeriesHead [
-
-	| actual expected |
-
-	expected := #(5.1 4.9 4.7 7 6.4) asDataSeries.
-	expected name: series name.
-	expected keys: (1 to: series defaultHeadTailSize).
-
-	actual := series head.
-
-	self assert: actual equals: expected.
-]
-
-{ #category : #tests }
-DataFrameHeadTailTest >> testDataSeriesHeadN [
-
-	| actual expected |
-
-	expected := #(5.1 4.9 4.7) asDataSeries.
-	expected name: series name.
-	expected keys: (1 to: 3).
-
-	actual := series head: 3.
-
-	self assert: actual equals: expected.
-]
-
-{ #category : #tests }
-DataFrameHeadTailTest >> testDataSeriesTail [
-
-	| actual expected |
-
-	expected := #(6.4 6.9 6.3 5.8 7.1) asDataSeries.
-	expected name: series name.
-	expected keys:
-		(series size - series defaultHeadTailSize + 1 to: series size).
-
-	actual := series tail.
-
-	self assert: actual equals: expected.
-]
-
-{ #category : #tests }
-DataFrameHeadTailTest >> testDataSeriesTailN [
-
-	| actual expected |
-
-	expected := #(6.3 5.8 7.1) asDataSeries.
-	expected name: series name.
-	expected keys:
-		(series size - 3 + 1 to: series size).
-
-	actual := series tail: 3.
-
-	self assert: actual equals: expected.
-]
-
 { #category : #tests }
 DataFrameHeadTailTest >> testDefaultHeadTailSize [
 

src/DataFrame-Tests/DataFrameQueriesTest.class.st

@@ -44,7 +44,7 @@ DataFrameQueriesTest >> testSelect [
 
 	actual := df
 		select: [ :row |
-			(row atKey: #a) = 'x' and: (row atKey: #b) < 30 ].
+			(row at: #a) = 'x' and: (row at: #b) < 30 ].
 
 	expected := DataFrame withRows: #(
 		(x 10 0.25 0.1)

src/DataFrame-Tests/DataFrameStatsTest.class.st

@@ -11,7 +11,7 @@ Class {
 DataFrameStatsTest >> setUp [
 
 	df := DataFrame withRows: #(
-		(5.1 3.5 1.4 0.2)
+		(7.1 3.5 1.4 0.2)
 		(4.9 3 1.4 0.2)
 		(4.7 3.2 1.3 0.2)
 		(7 3.2 4.7 1.4)
@@ -29,7 +29,7 @@ DataFrameStatsTest >> testAverage [
 
 	| expected actual |
 
-	expected := { 6.022222 . 3.133333 . 3.911111 . 1.277778 } asDataSeries.
+	expected := { 6.244444 . 3.133333 . 3.911111 . 1.277778 } asDataSeries.
 	expected name: #average.
 	expected keys: df columnNames.
 
@@ -42,7 +42,7 @@ DataFrameStatsTest >> testFirstQuartile [
 
 	| expected actual |
 
-	expected := { 5.1 . 3 . 1.4 . 0.2 } asDataSeries.
+	expected := { 5.8 . 3 . 1.4 . 0.2 } asDataSeries.
 	expected name: #firstQuartile.
 	expected keys: df columnNames.
 
@@ -55,7 +55,7 @@ DataFrameStatsTest >> testInterquartileRange [
 
 	| expected actual |
 
-	expected := { 1.8 . 0.2 . 3.7 . 1.7 } asDataSeries.
+	expected := { 1.2 . 0.2 . 3.7 . 1.7 } asDataSeries.
 	expected name: #interquartileRange.
 	expected keys: df columnNames.
 
@@ -81,7 +81,7 @@ DataFrameStatsTest >> testMedian [
 
 	| expected actual |
 
-	expected := { 6.3 . 3.2 . 4.7 . 1.5 } asDataSeries.
+	expected := { 6.4 . 3.2 . 4.7 . 1.5 } asDataSeries.
 	expected name: #median.
 	expected keys: df columnNames.
 
@@ -107,7 +107,7 @@ DataFrameStatsTest >> testMode [
 
 	| expected actual |
 
-	expected := { 4.7 . 3.2 . 1.4 . 0.2 } asDataSeries.
+	expected := { 7.1 . 3.2 . 1.4 . 0.2 } asDataSeries.
 	expected name: #mode.
 	expected keys: df columnNames.
 
@@ -133,7 +133,7 @@ DataFrameStatsTest >> testStdev [
 
 	| expected actual |
 
-	expected := { 0.936453 . 0.223607 . 1.971956 . 0.877180 } asDataSeries.
+	expected := { 0.927512 . 0.223607 . 1.971956 . 0.877180 } asDataSeries.
 
 	expected name: #stdev.
 	expected keys: df columnNames.
@@ -147,7 +147,7 @@ DataFrameStatsTest >> testThirdQuartile [
 
 	| expected actual |
 
-	expected := { 6.9 . 3.2 . 5.1 . 1.9 } asDataSeries.
+	expected := { 7 . 3.2 . 5.1 . 1.9 } asDataSeries.
 	expected name: #thirdQuartile.
 	expected keys: df columnNames.
 
@@ -160,7 +160,7 @@ DataFrameStatsTest >> testVariance [
 
 	| expected actual |
 
-	expected := { 0.876944 . 0.050000 . 3.888611 . 0.769444 } asDataSeries.
+	expected := { 0.860278 . 0.050000 . 3.888611 . 0.769444 } asDataSeries.
 	expected name: #variance.
 	expected keys: df columnNames.