Merge pull request #184 from jecisc/add-way-to-categorize-by-bins

Add a way to categorize a data series by bins

Author: jecisc

src/DataFrame-Tests/DataSeriesTest.class.st

@@ -1248,6 +1248,37 @@ DataSeriesTest >> testGroupByAggregateUsingSizeMismatch [
 		raise: SizeMismatch.
 ]
 
+{ #category : #grouping }
+DataSeriesTest >> testGroupByBins [
+
+	| actual expected |
+	actual := series groupByBins: { 0 . 5 . 10 . 15 . Float infinity }.
+	expected := DataSeries withKeys: keyArray values: #( 1 2 2 4 2 2 2 2 3 3 4 ) name: 'ExampleSeries'.
+	self assert: actual equals: expected
+]
+
+{ #category : #grouping }
+DataSeriesTest >> testGroupByBinsLabelled [
+
+	| actual expected |
+	actual := series
+		          groupByBins: { 0 . 5 . 10 . 15 . Float infinity }
+		          labelled: #( 0 5 10 15 ).
+	expected := DataSeries withKeys: keyArray values: #( 0 5 5 15 5 5 5 5 10 10 15 ) name: 'ExampleSeries'.
+	self assert: actual equals: expected
+]
+
+{ #category : #grouping }
+DataSeriesTest >> testGroupByBinsLabelledWithSizeProblem [
+
+	self
+		should: [
+			series
+				groupByBins: { 0 . 5 . 10 . 15 . Float infinity }
+				labelled: #( 0 5 10 ) ]
+		raise: SizeMismatch
+]
+
 { #category : #grouping }
 DataSeriesTest >> testGroupByUniqueValuesAndAggregateUsing [
 

src/DataFrame/DataSeries.class.st

@@ -326,6 +326,29 @@ DataSeries >> groupBy: otherSeries aggregateUsing: aBlock as: aNewName [
 	^ self class withKeys: groupMap keys values: (groupMap values collect: aBlock) name: aNewName
 ]
 
+{ #category : #grouping }
+DataSeries >> groupByBins: bins [
+
+	^ self groupByBins: bins labelled: (1 to: bins size - 1)
+]
+
+{ #category : #grouping }
+DataSeries >> groupByBins: bins labelled: aCollection [
+	"I receive two parameters:
+	- A collection of bins that will determine intervals to group the values
+	- A collection of labels to apply for each intervals of the bins
+
+	I return a new DataSeries associating each key to a label corresponding to the bin they match."
+
+	| labelledIntervals |
+	bins size = (aCollection size + 1) ifFalse: [ SizeMismatch signal: 'The labels should have one less elements than the bins.' ].
+
+	labelledIntervals := OrderedDictionary new.
+	bins overlappingPairsWithIndexDo: [ :min :max :index | labelledIntervals at: (aCollection at: index) put: min -> max ].
+
+	^ self collect: [ :each | labelledIntervals keyAtValue: (labelledIntervals values detect: [ :asso | each between: asso key and: asso value ]) ]
+]
+
 { #category : #grouping }
 DataSeries >> groupByUniqueValuesAndAggregateUsing: aBlock [
 	"Group my values by their unique values and aggregate them using aBlock. Use my name by default"