refactor: Train err and new Test

lvisei · Dec 15, 2020 · b501d03 · b501d03
1 parent 12fd1bb
commit b501d03
Show file tree

Hide file tree

Showing 5 changed files with 72 additions and 9 deletions.
diff --git a/examples/csv/main.go b/examples/csv/main.go
@@ -51,7 +51,9 @@ func main() {
 	defer timeCost()("训练模型与插值生成网格图片总耗时")
 
 	ordinaryKriging := ordinarykriging.NewOrdinary(data["values"], data["x"], data["y"])
-	_ = ordinaryKriging.Train(ordinarykriging.Spherical, 0, 100)
+	if _, err := ordinaryKriging.Train(ordinarykriging.Exponential, 0, 100); err != nil {
+		log.Fatal(err)
+	}
 
 	_ = polygon
 	gridPlot(ordinaryKriging, polygon)

diff --git a/examples/tinygo/main.go b/examples/tinygo/main.go
@@ -2,6 +2,7 @@ package main
 
 import (
 	"github.com/liuvigongzuoshi/go-kriging/ordinarykriging"
+	"log"
 )
 
 // tinygo build -o kriging.wasm -opt z -heap-size 2048M  -target wasm ./main.go
@@ -1832,12 +1833,17 @@ func OrdinaryKrigingFunc() int {
 
 func OrdinaryKriging(values, lons, lats []float64, model string, sigma2, alpha, width float64, polygon ordinarykriging.PolygonCoordinates) *ordinarykriging.GridMatrices {
 	ordinaryKriging := ordinarykriging.NewOrdinary(values, lons, lats)
-	_ = ordinaryKriging.Train(ordinarykriging.ModelType(model), sigma2, alpha)
+	if _, err := ordinaryKriging.Train(ordinarykriging.ModelType(model), sigma2, alpha); err != nil {
+		log.Fatal(err)
+	}
 	return ordinaryKriging.Grid(polygon, width)
 }
 
 func OrdinaryKrigingTrain(values, lons, lats []float64, model string, sigma2 float64, alpha float64) *ordinarykriging.Variogram {
 	ordinaryKriging := ordinarykriging.NewOrdinary(values, lons, lats)
-	variogram := ordinaryKriging.Train(ordinarykriging.ModelType(model), sigma2, alpha)
+	variogram, err := ordinaryKriging.Train(ordinarykriging.ModelType(model), sigma2, alpha)
+	if err != nil {
+		log.Fatal(err)
+	}
 	return variogram
 }
diff --git a/examples/wasm/main.go b/examples/wasm/main.go
@@ -85,13 +85,18 @@ func OrdinaryKrigingTrainFunc(this js.Value, args []js.Value) interface{} {
 
 func RunOrdinaryKrigingTrain(values, lons, lats []float64, model string, sigma2 float64, alpha float64) *ordinarykriging.Variogram {
 	ordinaryKriging := ordinarykriging.NewOrdinary(values, lons, lats)
-	variogram := ordinaryKriging.Train(ordinarykriging.ModelType(model), sigma2, alpha)
+	variogram, err := ordinaryKriging.Train(ordinarykriging.ModelType(model), sigma2, alpha)
+	if err != nil {
+		log.Fatal(err)
+	}
 	return variogram
 }
 
 func RunOrdinaryKriging(values, lons, lats []float64, model string, sigma2, alpha, width float64, polygon ordinarykriging.PolygonCoordinates) *ordinarykriging.GridMatrices {
 	ordinaryKriging := ordinarykriging.NewOrdinary(values, lons, lats)
-	_ = ordinaryKriging.Train(ordinarykriging.ModelType(model), sigma2, alpha)
+	if _, err := ordinaryKriging.Train(ordinarykriging.ModelType(model), sigma2, alpha); err != nil {
+		log.Fatal(err)
+	}
 	return ordinaryKriging.Grid(polygon, width)
 }
 

diff --git a/ordinarykriging/ordinarykriging.go b/ordinarykriging/ordinarykriging.go
@@ -5,6 +5,7 @@
 package ordinarykriging
 
 import (
+	"errors"
 	"github.com/liuvigongzuoshi/go-kriging/canvas"
 	"image"
 	"image/color"
@@ -62,7 +63,7 @@ func krigingVariogramSpherical(h, nugget, range_, sill, A float64) float64 {
 }
 
 // Train using gaussian processes with bayesian priors
-func (variogram *Variogram) Train(model ModelType, sigma2 float64, alpha float64) *Variogram {
+func (variogram *Variogram) Train(model ModelType, sigma2 float64, alpha float64) (*Variogram, error) {
 	variogram.Nugget = 0.0
 	variogram.Range = 0.0
 	variogram.Sill = 0.0
@@ -147,8 +148,7 @@ func (variogram *Variogram) Train(model ModelType, sigma2 float64, alpha float64
 			k = 0
 		}
 		if l < 2 {
-			// Error: Not enough points
-			return variogram
+			return nil, errors.New("not enough points")
 		}
 	}
 
@@ -232,7 +232,7 @@ func (variogram *Variogram) Train(model ModelType, sigma2 float64, alpha float64
 	variogram.K = K
 	variogram.M = M
 
-	return variogram
+	return variogram, nil
 }
 
 // Predict model prediction

diff --git a/ordinarykriging/ordinarykriging_test.go b/ordinarykriging/ordinarykriging_test.go
@@ -36,6 +36,56 @@ func generateData(count int) (FloatList, FloatList, FloatList) {
 	return randomValues, randomLats, randomLons
 }
 
+func TestVariogram_Predict_100(t *testing.T) {
+	var randomValues, randomLats, randomLons FloatList
+	randomLons = append(randomLons, randFloats(-90, 90, 100)...)
+	randomLats = append(randomLats, randFloats(-180, 180, 100)...)
+	for _, lon := range randomLons {
+		var value float64
+		if lon > 0 {
+			value = 100
+		} else {
+			value = 0
+		}
+		randomValues = append(randomValues, value)
+	}
+	ordinaryKriging := ordinarykriging.NewOrdinary(randomValues, randomLats, randomLons)
+	if _, err := ordinaryKriging.Train(ordinarykriging.Exponential, 0, 10); err != nil {
+		t.Fatal("variogram is null", err)
+	}
+	if ordinaryKriging.Predict(180, 0) < 50 {
+		t.Fatal("unexpected result (<50)")
+	}
+	if ordinaryKriging.Predict(-180, 0) > 50 {
+		t.Fatal("unexpected result (>50)")
+	}
+}
+
+func TestVariogram_Predict_1000(t *testing.T) {
+	var randomValues, randomLats, randomLons FloatList
+	randomLons = append(randomLons, randFloats(-90, 90, 1000)...)
+	randomLats = append(randomLats, randFloats(-180, 180, 1000)...)
+	for _, lon := range randomLons {
+		var value float64
+		if lon > 0 {
+			value = 100
+		} else {
+			value = 0
+		}
+		randomValues = append(randomValues, value)
+	}
+	ordinaryKriging := ordinarykriging.NewOrdinary(randomValues, randomLats, randomLons)
+	if _, err := ordinaryKriging.Train(ordinarykriging.Exponential, 0, 10); err != nil {
+		t.Fatal("variogram is null", err)
+	}
+	if ordinaryKriging.Predict(180, 0) < 50 {
+		t.Fatal("unexpected result (<50)")
+	}
+	if ordinaryKriging.Predict(-180, 0) > 50 {
+		t.Fatal("unexpected result (>50)")
+	}
+}
+
 func TestVariogram_Plot(t *testing.T) {
 	ordinaryKriging := ordinarykriging.NewOrdinary(randomValues, randomLats, randomLons)
 	ordinaryKriging.Train(ordinarykriging.Exponential, 0, 100)