Optimizing Loss Calculation

examples/trees/cart.go

@@ -35,10 +35,13 @@ func main() {
 
 	// Create New Classification Tree
 	// Hyperparameters - loss function, max Depth (-1 will split until pure), list of unique labels
-	decTree = NewDecisionTreeClassifier("entropy", -1, []int64{0, 1})
+	decTree := NewDecisionTreeClassifier("entropy", -1, []int64{0, 1})
 
 	// Train Tree
-	decTree.Fit(trainData)
+	err = decTree.Fit(trainData)
+	if err != nil {
+		panic(err)
+	}
 	// Print out tree for visualization - shows splits and feature and predictions
 	fmt.Println(decTree.String())
 
@@ -62,7 +65,10 @@ func main() {
 	regTree := NewDecisionTreeRegressor("mse", -1)
 
 	// Train Tree
-	regTree.Fit(trainRegData)
+	err = regTree.Fit(trainRegData)
+	if err != nil {
+		panic(err)
+	}
 
 	// Print out tree for visualization
 	fmt.Println(regTree.String())

trees/cart_classifier.go

@@ -39,25 +39,31 @@ type CARTDecisionTreeClassifier struct {
 	triedSplits [][]float64
 }
 
+// Convert a series of labels to frequency map for efficient impurity calculation
+func convertToMap(y []int64, labels []int64) map[int64]int {
+	labelCount := make(map[int64]int)
+	for _, label := range labels {
+		labelCount[label] = 0
+	}
+	for _, value := range y {
+		labelCount[value]++
+	}
+	return labelCount
+}
+
 // Calculate Gini Impurity of Target Labels
 func computeGiniImpurityAndModeLabel(y []int64, labels []int64) (float64, int64) {
 	nInstances := len(y)
 	gini := 0.0
-	maxLabelCount := 0
 	var maxLabel int64 = 0
-	for label := range labels {
-		numLabel := 0
-		for target := range y {
-			if y[target] == labels[label] {
-				numLabel++
+
+	labelCount := convertToMap(y, labels)
+	for _, label := range labels {
+		if labelCount[label] > labelCount[maxLabel] {
+			maxLabel = label
 		}
-		}
-		p := float64(numLabel) / float64(nInstances)
+		p := float64(labelCount[label]) / float64(nInstances)
 		gini += p * (1 - p)
-		if numLabel > maxLabelCount {
-			maxLabel = labels[label]
-			maxLabelCount = numLabel
-		}
 	}
 	return gini, maxLabel
 }
@@ -66,26 +72,19 @@ func computeGiniImpurityAndModeLabel(y []int64, labels []int64) (float64, int64)
 func computeEntropyAndModeLabel(y []int64, labels []int64) (float64, int64) {
 	nInstances := len(y)
 	entropy := 0.0
-	maxLabelCount := 0
 	var maxLabel int64 = 0
-	for label := range labels {
-		numLabel := 0
-		for target := range y {
-			if y[target] == labels[label] {
-				numLabel++
-			}
-		}
-		p := float64(numLabel) / float64(nInstances)
 
+	labelCount := convertToMap(y, labels)
+	for _, label := range labels {
+		if labelCount[label] > labelCount[maxLabel] {
+			maxLabel = label
+		}
+		p := float64(labelCount[label]) / float64(nInstances)
 		logP := math.Log2(p)
 		if p == 0 {
 			logP = 0
 		}
-		entropy += -p * logP
-		if numLabel > maxLabelCount {
-			maxLabel = labels[label]
-			maxLabelCount = numLabel
-		}
+		entropy += (-p * logP)
 	}
 	return entropy, maxLabel
 }

trees/cart_test.go

@@ -1,7 +1,6 @@
 package trees
 
 import (
-	"fmt"
 	"testing"
 
 	. "github.com/smartystreets/goconvey/convey"
@@ -42,8 +41,7 @@ func TestRegressor(t *testing.T) {
 
 		// is data reordered correctly
 		orderedData, orderedY := classifierReOrderData(getFeature(classifierData, 1), classifierData, classifiery)
-		fmt.Println(orderedData)
-		fmt.Println(orderedY)
+
 		So(orderedData[1][1], ShouldEqual, 3.0)
 		So(orderedY[0], ShouldEqual, 1)
 
@@ -81,9 +79,9 @@ func TestRegressor(t *testing.T) {
 		leftData, rightData, leftY, rightY := regressorCreateSplit(data, 1, y, 5.0)
 
 		So(len(leftData), ShouldEqual, 2)
-		So(len(lefty), ShouldEqual, 2)
+		So(len(leftY), ShouldEqual, 2)
 		So(len(rightData), ShouldEqual, 2)
-		So(len(righty), ShouldEqual, 2)
+		So(len(rightY), ShouldEqual, 2)
 
 		// is data reordered correctly
 		regressorOrderedData, regressorOrderedY := regressorReOrderData(getFeature(data, 1), data, y)

trees/cart_utils.go

@@ -4,23 +4,14 @@ import (
 	"github.com/sjwhitworth/golearn/base"
 )
 
-// Helper Function to check if data point is unique or not.
-// We will use this to isolate unique values of a feature
-func stringInSlice(a float64, list []float64) bool {
-	for _, b := range list {
-		if b == a {
-			return true
-		}
-	}
-	return false
-}
-
 // Isolate only unique values. This way, we can try only unique splits and not redundant ones.
 func findUnique(data []float64) []float64 {
-	var unique []float64
-	for i := range data {
-		if !stringInSlice(data[i], unique) {
-			unique = append(unique, data[i])
+	keys := make(map[float64]bool)
+	unique := []float64{}
+	for _, entry := range data {
+		if _, value := keys[entry]; !value {
+			keys[entry] = true
+			unique = append(unique, entry)
 		}
 	}
 	return unique