Implemented groupBy for DataSeries

olekscode · olekscode · commit 80ff875f5980 · 2017-07-16T20:10:54.000+03:00
diff --git a/DataFrame-Core.package/DataFrame.class/class/fromCsv.separatedBy..st b/DataFrame-Core.package/DataFrame.class/class/fromCsv.separatedBy..st
@@ -9,7 +9,7 @@ fromCsv: pathToFile separatedBy: separator
 		reject: [:e | e isEmpty]) " lose empty lines "
 		collect: [:e | e findTokens: separator]. " split into fields "
 
-	df := self rows: (data copyFrom: 2 to: data size).	
+	df := self fromRows: (data copyFrom: 2 to: data size).	
 	df columnNames: (data at: 1) asArray.
 	
 	"Deduce types"
diff --git a/DataFrame-Core.package/DataFrameGrouped.class/README.md b/DataFrame-Core.package/DataFrameGrouped.class/README.md
diff --git a/DataFrame-Core.package/DataFrameGrouped.class/instance/apply..st b/DataFrame-Core.package/DataFrameGrouped.class/instance/apply..st
@@ -0,0 +1,4 @@
+private
+apply: aBlock
+	
+	"TODO"
diff --git a/DataFrame-Core.package/DataFrameGrouped.class/properties.json b/DataFrame-Core.package/DataFrameGrouped.class/properties.json
@@ -0,0 +1,11 @@
+{
+	"commentStamp" : "",
+	"super" : "DataGrouped",
+	"category" : "DataFrame-Core",
+	"classinstvars" : [ ],
+	"pools" : [ ],
+	"classvars" : [ ],
+	"instvars" : [ ],
+	"name" : "DataFrameGrouped",
+	"type" : "normal"
+}
diff --git a/DataFrame-Core.package/DataGrouped.class/README.md b/DataFrame-Core.package/DataGrouped.class/README.md
diff --git a/DataFrame-Core.package/DataGrouped.class/instance/apply..st b/DataFrame-Core.package/DataGrouped.class/instance/apply..st
@@ -0,0 +1,4 @@
+private
+apply: aBlock
+
+	self subclassResponsibility
diff --git a/DataFrame-Core.package/DataGrouped.class/instance/average.st b/DataFrame-Core.package/DataGrouped.class/instance/average.st
@@ -0,0 +1,4 @@
+private
+average
+
+	^ self apply: [ :eachGroup | eachGroup average ]
diff --git a/DataFrame-Core.package/DataGrouped.class/instance/count.st b/DataFrame-Core.package/DataGrouped.class/instance/count.st
@@ -0,0 +1,4 @@
+private
+count
+
+	^ self apply: [ :eachGroup | eachGroup size ]
diff --git a/DataFrame-Core.package/DataGrouped.class/instance/max.st b/DataFrame-Core.package/DataGrouped.class/instance/max.st
@@ -0,0 +1,4 @@
+private
+max
+
+	^ self apply: [ :eachGroup | eachGroup max ]
diff --git a/DataFrame-Core.package/DataGrouped.class/instance/min.st b/DataFrame-Core.package/DataGrouped.class/instance/min.st
@@ -0,0 +1,4 @@
+private
+min
+
+	^ self apply: [ :eachGroup | eachGroup min ]
diff --git a/DataFrame-Core.package/DataGrouped.class/instance/sum.st b/DataFrame-Core.package/DataGrouped.class/instance/sum.st
@@ -0,0 +1,4 @@
+private
+sum
+
+	^ self apply: [ :eachGroup | eachGroup sum ]
diff --git a/DataFrame-Core.package/DataGrouped.class/properties.json b/DataFrame-Core.package/DataGrouped.class/properties.json
@@ -0,0 +1,11 @@
+{
+	"commentStamp" : "",
+	"super" : "Object",
+	"category" : "DataFrame-Core",
+	"classinstvars" : [ ],
+	"pools" : [ ],
+	"classvars" : [ ],
+	"instvars" : [ ],
+	"name" : "DataGrouped",
+	"type" : "normal"
+}
diff --git a/DataFrame-Core.package/DataSeries.class/instance/groupBy..st b/DataFrame-Core.package/DataSeries.class/instance/groupBy..st
@@ -0,0 +1,4 @@
+as yet unclassified
+groupBy: aSeries
+
+	^ DataSeriesGrouped group: self by: aSeries
diff --git a/DataFrame-Core.package/DataSeries.class/instance/initialize.st b/DataFrame-Core.package/DataSeries.class/instance/initialize.st
@@ -2,4 +2,4 @@ instance creation
 initialize
 
 	super initialize.
-	keys := Array new.
+	keys := (1 to: self size) asArray.
diff --git a/DataFrame-Core.package/DataSeries.class/instance/printOn..st b/DataFrame-Core.package/DataSeries.class/instance/printOn..st
@@ -1,14 +1,19 @@
 printing
 printOn: aStream
 
-	self name isNil
-		ifTrue: [ aStream nextPutAll: '(unnamed)' ]
-		ifFalse: [ aStream nextPutAll: '[', self name asString, ']' ].
-		
+	aStream nextPutAll: 'a '.
+	self class printOn: aStream.
 	aStream cr.
+
+	self name isNil
+		ifFalse: [
+			aStream nextPutAll: '''', self name asString, ''''.
+			aStream cr ].
 	
-	self keysAndValuesDo: [ :key :value |
-		aStream nextPutAll: key asString.
+	self withIndexDo: [ :each :i |
+		aStream nextPutAll: (self keys at: i) asString.
 		aStream tab.
-		aStream nextPutAll: value asString.
-		aStream cr ].
+		each printOn: aStream.
+		
+		i = self size 
+			ifFalse: [ aStream cr ] ].
diff --git a/DataFrame-Core.package/DataSeriesGrouped.class/README.md b/DataFrame-Core.package/DataSeriesGrouped.class/README.md
diff --git a/DataFrame-Core.package/DataSeriesGrouped.class/class/group.by..st b/DataFrame-Core.package/DataSeriesGrouped.class/class/group.by..st
@@ -0,0 +1,4 @@
+as yet unclassified
+group: firstSeries by: secondSeries
+
+	^ self new split: firstSeries by: secondSeries
diff --git a/DataFrame-Core.package/DataSeriesGrouped.class/instance/apply..st b/DataFrame-Core.package/DataSeriesGrouped.class/instance/apply..st
@@ -0,0 +1,12 @@
+private
+apply: aBlock
+
+	| result |
+	
+	result := groups collect: [ :eachGroup |
+		aBlock value: eachGroup ].
+	
+	result := result asDataSeries.
+	result keys: groups keys.
+	
+	^ result
diff --git a/DataFrame-Core.package/DataSeriesGrouped.class/instance/printOn..st b/DataFrame-Core.package/DataSeriesGrouped.class/instance/printOn..st
@@ -0,0 +1,21 @@
+printing
+printOn: aStream
+
+	super printOn: aStream.
+	aStream cr.
+
+	groups doWithIndex: [ :eachGroup :i |
+		aStream nextPutAll: (groups keys at: i) asString.
+		aStream tab.
+		aStream nextPutAll: '('.
+		
+		eachGroup doWithIndex: [ :each :j |
+			each printOn: aStream.
+			
+			j = eachGroup size
+				ifFalse: [ aStream space ] ].
+		
+		aStream nextPutAll: ')'.
+		
+		i = groups size
+			ifFalse: [ aStream cr ] ]
diff --git a/DataFrame-Core.package/DataSeriesGrouped.class/instance/split.by..st b/DataFrame-Core.package/DataSeriesGrouped.class/instance/split.by..st
@@ -0,0 +1,24 @@
+private
+split: firstSeries by: secondSeries
+
+	| secondUnique |
+	
+	firstSeries size = secondSeries size
+		ifFalse: [ SizeMismatch signal ].
+	
+	secondUnique := secondSeries unique asArray.
+	
+	groups := secondUnique collect: [ :eachUnique |
+		| aList |
+		aList := LinkedList new.
+		
+		secondSeries doWithIndex: [ :eachSecond :i |
+			eachSecond = eachUnique 
+				ifTrue: [ aList add: (firstSeries at: i) ] ].
+		
+		aList asDataSeries ].
+	
+	groups := groups asDataSeries.
+	groups keys: secondUnique.
+	
+	^ self
diff --git a/DataFrame-Core.package/DataSeriesGrouped.class/properties.json b/DataFrame-Core.package/DataSeriesGrouped.class/properties.json
@@ -0,0 +1,13 @@
+{
+	"commentStamp" : "",
+	"super" : "DataGrouped",
+	"category" : "DataFrame-Core",
+	"classinstvars" : [ ],
+	"pools" : [ ],
+	"classvars" : [ ],
+	"instvars" : [
+		"groups"
+	],
+	"name" : "DataSeriesGrouped",
+	"type" : "normal"
+}

-Original file line number
+Diff line change
@@ @@ -0,0 +1,4 @@ @@
 +private
 +apply: aBlock
++
 +	"TODO"