v0.0.482 mathext.Percentile

mathext/statistics.go

@@ -1,6 +1,9 @@
 package mathext
 
-import "gogs.mikescher.com/BlackForestBytes/goext/langext"
+import (
+	"gogs.mikescher.com/BlackForestBytes/goext/exerr"
+	"gogs.mikescher.com/BlackForestBytes/goext/langext"
+)
 
 func Sum[T langext.NumberConstraint](v []T) T {
 	total := T(0)
@@ -41,3 +44,53 @@ func ArrMax[T langext.OrderedConstraint](v []T) T {
 	}
 	return r
 }
+
+func MustPercentile[T langext.NumberConstraint](rawdata []T, percentile float64) T {
+	v, err := Percentile(rawdata, percentile)
+	if err != nil {
+		panic(err)
+	}
+	return v
+}
+
+func Percentile[T langext.NumberConstraint](rawdata []T, percentile float64) (T, error) {
+	v, err := FloatPercentile(rawdata, percentile)
+	if err != nil {
+		return T(0), err
+	}
+	return T(v), nil
+
+}
+
+func FloatPercentile[T langext.NumberConstraint](rawdata []T, percentile float64) (float64, error) {
+	if len(rawdata) == 0 {
+		return 0, exerr.New(exerr.TypeAssert, "no data to calculate percentile").Any("percentile", percentile).Build()
+	}
+
+	if percentile < 0 || percentile > 100 {
+		return 0, exerr.New(exerr.TypeAssert, "percentile out of range").Any("percentile", percentile).Build()
+	}
+
+	data := langext.ArrCopy(rawdata)
+	langext.Sort(data)
+
+	idxFloat := float64(len(data)-1) * (percentile / float64(100))
+
+	idxInt := int(idxFloat)
+
+	// exact match on index
+	if idxFloat == float64(idxInt) {
+		return float64(data[idxInt]), nil
+	}
+
+	// linear interpolation
+	v1 := data[idxInt]
+	v2 := data[idxInt+1]
+
+	weight := idxFloat - float64(idxInt)
+
+	valFloat := (float64(v1) * (1 - weight)) + (float64(v2) * weight)
+
+	return valFloat, nil
+
+}

mathext/statistics_test.go

@@ -0,0 +1,238 @@
+package mathext
+
+import (
+	"math"
+	"testing"
+)
+
+func TestSumIntsHappyPath(t *testing.T) {
+	values := []int{1, 2, 3, 4, 5}
+	expected := 15
+	result := Sum(values)
+	if result != expected {
+		t.Errorf("Sum of %v; expected %v, got %v", values, expected, result)
+	}
+}
+
+func TestSumFloatsHappyPath(t *testing.T) {
+	values := []float64{1.1, 2.2, 3.3}
+	expected := 6.6
+	result := Sum(values)
+	if result != expected {
+		t.Errorf("Sum of %v; expected %v, got %v", values, expected, result)
+	}
+}
+
+func TestMeanOfInts(t *testing.T) {
+	values := []float64{1, 2, 3, 4, 5}
+	expected := 3.0
+	result := Mean(values)
+	if result != expected {
+		t.Errorf("Mean of %v; expected %v, got %v", values, expected, result)
+	}
+}
+
+func TestMedianOddNumberOfElements(t *testing.T) {
+	values := []float64{1, 2, 3, 4, 5}
+	expected := 3.0
+	result := Median(values)
+	if result != expected {
+		t.Errorf("Median of %v; expected %v, got %v", values, expected, result)
+	}
+}
+
+func TestMedianEvenNumberOfElements(t *testing.T) {
+	values := []float64{1, 2, 3, 4, 5, 6}
+	expected := 3.5
+	result := Median(values)
+	if result != expected {
+		t.Errorf("Median of %v; expected %v, got %v", values, expected, result)
+	}
+}
+
+func TestArrMinInts(t *testing.T) {
+	values := []int{5, 3, 9, 1, 4}
+	expected := 1
+	result := ArrMin(values)
+	if result != expected {
+		t.Errorf("ArrMin of %v; expected %v, got %v", values, expected, result)
+	}
+}
+
+func TestArrMaxInts(t *testing.T) {
+	values := []int{5, 3, 9, 1, 4}
+	expected := 9
+	result := ArrMax(values)
+	if result != expected {
+		t.Errorf("ArrMax of %v; expected %v, got %v", values, expected, result)
+	}
+}
+
+func TestPercentileValidInput(t *testing.T) {
+	values := []int{1, 2, 3, 4, 5}
+	percentile := 50.0
+	expected := 3
+	result, err := Percentile(values, percentile)
+	if err != nil || result != expected {
+		t.Errorf("Percentile %v of %v; expected %v, got %v, err: %v", percentile, values, expected, result, err)
+	}
+}
+
+func TestPercentileOutOfRange(t *testing.T) {
+	values := []int{1, 2, 3, 4, 5}
+	percentile := 150.0
+	_, err := Percentile(values, percentile)
+	if err == nil {
+		t.Errorf("Expected error for percentile %v out of range, got nil", percentile)
+	}
+}
+
+func TestPercentileValueInArray(t *testing.T) {
+	values := []int{1, 3, 5, 7, 9}
+	percentile := 40.0
+	expected := 4
+	result, err := Percentile(values, percentile)
+	if err != nil || result != expected {
+		t.Errorf("Percentile %v of %v; expected %v, got %v, err: %v", percentile, values, expected, result, err)
+	}
+}
+
+func TestFloatPercentileValueInArray(t *testing.T) {
+	values := []int{1, 3, 5, 7, 9}
+	percentile := 40.0
+	expected := 4.2
+	result, err := FloatPercentile(values, percentile)
+	if err != nil || result != expected {
+		t.Errorf("Percentile %v of %v; expected %v, got %v, err: %v", percentile, values, expected, result, err)
+	}
+}
+
+func TestPercentileInterpolation(t *testing.T) {
+	values := []float64{1.0, 2.0, 3.0, 4.0, 5.0}
+	percentile := 25.0
+	expected := 2.0
+	result, err := Percentile(values, percentile)
+	if err != nil || result != expected {
+		t.Errorf("Percentile %v of %v; expected %v, got %v, err: %v", percentile, values, expected, result, err)
+	}
+}
+
+func TestPercentileSingleValue(t *testing.T) {
+	values := []int{10}
+	percentile := 50.0
+	expected := 10
+	result, err := Percentile(values, percentile)
+	if err != nil || result != expected {
+		t.Errorf("Percentile %v of %v; expected %v, got %v, err: %v", percentile, values, expected, result, err)
+	}
+}
+
+func TestPercentileExactlyBetweenTwoValues(t *testing.T) {
+	values := []float64{1, 2, 3, 4, 5}
+	percentile := 62.5 // Exactly between 3 and 4
+	expected := 3.5
+	result, err := FloatPercentile(values, percentile)
+	if err != nil || result != expected {
+		t.Errorf("Percentile %v of %v; expected %v, got %v, err: %v", percentile, values, expected, result, err)
+	}
+}
+
+func TestPercentileTwoThirdsBetweenTwoValues(t *testing.T) {
+	values := []float64{0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10}
+	percentile := 66.666666666666
+	expected := 6.666666666666667 // Since 2/3 of the way between 6 and 7 is 6.666...
+	result, err := Percentile(values, percentile)
+	if err != nil || math.Abs(result-expected) > 1e-9 {
+		t.Errorf("Percentile %v of %v; expected %v, got %v, err: %v", percentile, values, expected, result, err)
+	}
+}
+
+func TestPercentileBetweenTwoValues1(t *testing.T) {
+	values := []float64{0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10}
+	percentile := 11.0
+	expected := 1.1
+	result, err := Percentile(values, percentile)
+	if err != nil || math.Abs(result-expected) > 1e-9 {
+		t.Errorf("Percentile %v of %v; expected %v, got %v, err: %v", percentile, values, expected, result, err)
+	}
+}
+
+func TestPercentileBetweenTwoValues2(t *testing.T) {
+	values := []float64{0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10}
+	percentile := 9.0
+	expected := 0.9
+	result, err := Percentile(values, percentile)
+	if err != nil || math.Abs(result-expected) > 1e-9 {
+		t.Errorf("Percentile %v of %v; expected %v, got %v, err: %v", percentile, values, expected, result, err)
+	}
+}
+
+func TestPercentileUnsortedInput(t *testing.T) {
+	values := []float64{5, 1, 4, 2, 3} // Unsorted input
+	percentile := 50.0
+	expected := 3.0
+	result, err := Percentile(values, percentile)
+	if err != nil || result != expected {
+		t.Errorf("Percentile %v of %v; expected %v, got %v, err: %v", percentile, values, expected, result, err)
+	}
+}
+
+func TestPercentileUnsortedInputLowPercentile(t *testing.T) {
+	values := []float64{10, 6, 7, 3, 2, 9, 8, 1, 4, 5} // Unsorted input
+	percentile := 10.0
+	expected := 1.9 // Expecting interpolation between 1 and 2
+	result, err := Percentile(values, percentile)
+	if err != nil || result != expected {
+		t.Errorf("Percentile %v of %v; expected %v, got %v, err: %v", percentile, values, expected, result, err)
+	}
+}
+
+func TestPercentileUnsortedInputHighPercentile(t *testing.T) {
+	values := []float64{10, 6, 7, 3, 2, 9, 8, 1, 4, 5} // Unsorted input
+	percentile := 90.0
+	expected := 9.1 // Expecting interpolation between 9 and 10
+	result, err := Percentile(values, percentile)
+	if err != nil || result != expected {
+		t.Errorf("Percentile %v of %v; expected %v, got %v, err: %v", percentile, values, expected, result, err)
+	}
+}
+
+func TestFloatPercentileExactValueFromInput(t *testing.T) {
+	values := []float64{1.5, 2.5, 3.5, 4.5, 5.5}
+	percentile := 50.0 // Exact value from input array should be 3.5
+	expected := 3.5
+	result, err := FloatPercentile(values, percentile)
+	if err != nil || result != expected {
+		t.Errorf("FloatPercentile %v of %v; expected %v, got %v, err: %v", percentile, values, expected, result, err)
+	}
+}
+
+func TestFloatPercentileInterpolatedValue(t *testing.T) {
+	values := []float64{1.0, 2.0, 3.0, 4.0, 5.0}
+	percentile := 87.5 // Interpolated value between 4.0 and 5.0
+	expected := 4.5
+	result, err := FloatPercentile(values, percentile)
+	if err != nil || result != expected {
+		t.Errorf("FloatPercentile %v of %v; expected %v, got %v, err: %v", percentile, values, expected, result, err)
+	}
+}
+
+func TestFloatPercentileUnsortedInputExactValue(t *testing.T) {
+	values := []float64{5.5, 1.5, 4.5, 2.5, 3.5} // Unsorted input
+	percentile := 50.0
+	expected := 3.5
+	result, err := FloatPercentile(values, percentile)
+	if err != nil || result != expected {
+		t.Errorf("FloatPercentile %v of %v; expected %v, got %v, err: %v", percentile, values, expected, result, err)
+	}
+}
+
+func TestFloatPercentileUnsortedInputInterpolatedValue(t *testing.T) {
+	values := []float64{10.5, 6.5, 7.5, 3.5, 2.5, 9.5, 8.5, 1.5, 4.5, 5.5}
+	percentile := 80.0 // Interpolated value between 4.0 and 5.0
+	expected := 8.7
+	result, err := FloatPercentile(values, percentile)
+	if err != nil || math.Abs(result-expected) > 1e-9 {
+		t.Errorf("FloatPercentile %v of %v; expected %v, got %v, err: %v", percentile, values, expected, result, err)
+	}
+}