Add exponentially weighted moving average

gcc/ewma.go

@@ -0,0 +1,40 @@
+// SPDX-FileCopyrightText: 2025 The Pion community <https://pion.ly>
+// SPDX-License-Identifier: MIT
+
+package gcc
+
+type ewma struct {
+	initialized bool
+	alpha       float64
+	average     float64
+	variance    float64
+}
+
+func newEWMA(alpha float64) *ewma {
+	return &ewma{
+		initialized: false,
+		alpha:       alpha,
+		average:     0,
+		variance:    0,
+	}
+}
+
+func (a *ewma) update(sample float64) {
+	if !a.initialized {
+		a.initialized = true
+		a.average = sample
+
+		return
+	}
+	delta := sample - a.average
+	a.average += a.alpha * delta
+	a.variance = (1 - a.alpha) * (a.variance + a.alpha*delta*delta)
+}
+
+func (a *ewma) avg() float64 {
+	return a.average
+}
+
+func (a *ewma) varr() float64 {
+	return a.variance
+}

gcc/ewma_test.go

@@ -0,0 +1,124 @@
+// SPDX-FileCopyrightText: 2025 The Pion community <https://pion.ly>
+// SPDX-License-Identifier: MIT
+
+package gcc
+
+import (
+	"fmt"
+	"testing"
+
+	"github.com/stretchr/testify/assert"
+)
+
+// python to generate test cases:
+// import numpy as np
+// import pandas as pd
+// data = np.random.randint(1, 10, size=10)
+// df = pd.DataFrame(data)
+// expectedAvg = df.ewm(alpha=0.9, adjust=False).mean()
+// expectedVar = df.ewm(alpha=0.9, adjust=False).var(bias=True)
+
+func TestEWMA(t *testing.T) {
+	cases := []struct {
+		alpha       float64
+		samples     []float64
+		expectedAvg []float64
+		expectedVar []float64
+	}{
+		{
+			alpha:       0.9,
+			samples:     []float64{},
+			expectedAvg: []float64{},
+			expectedVar: []float64{},
+		},
+		{
+			alpha:       0.95,
+			samples:     []float64{0, 1, 2, 3, 4},
+			expectedAvg: []float64{0, 0.95, 1.9475, 2.947375, 3.947369},
+			expectedVar: []float64{0, 0.047500, 0.054744, 0.055356, 0.05539},
+		},
+		{
+			alpha:   0.9,
+			samples: []float64{1, 2, 3, 4},
+			expectedAvg: []float64{
+				1.000,
+				1.900,
+				2.890,
+				3.889,
+			},
+			expectedVar: []float64{
+				0.000000,
+				0.090000,
+				0.117900,
+				0.122679,
+			},
+		},
+		{
+			alpha:   0.9,
+			samples: []float64{8, 8, 5, 1, 3, 1, 8, 2, 8, 9},
+			expectedAvg: []float64{
+				8.000000,
+				8.000000,
+				5.300000,
+				1.430000,
+				2.843000,
+				1.184300,
+				7.318430,
+				2.531843,
+				7.453184,
+				8.845318,
+			},
+			expectedVar: []float64{
+				0.000000,
+				0.000000,
+				0.810000,
+				1.745100,
+				0.396351,
+				0.345334,
+				4.215372,
+				2.967250,
+				2.987792,
+				0.514117,
+			},
+		},
+		{
+			alpha:   0.9,
+			samples: []float64{7, 5, 6, 7, 3, 6, 8, 9, 5, 5},
+			expectedAvg: []float64{
+				7.000000,
+				5.200000,
+				5.920000,
+				6.892000,
+				3.389200,
+				5.738920,
+				7.773892,
+				8.877389,
+				5.387739,
+				5.038774,
+			},
+			expectedVar: []float64{
+				0.000000,
+				0.360000,
+				0.093600,
+				0.114336,
+				1.374723,
+				0.750937,
+				0.535217,
+				0.188822,
+				1.371955,
+				0.150726,
+			},
+		},
+	}
+
+	for i, tc := range cases {
+		t.Run(fmt.Sprintf("%v", i), func(t *testing.T) {
+			ewma := newEWMA(tc.alpha)
+			for i, sample := range tc.samples {
+				ewma.update(sample)
+				assert.InDelta(t, tc.expectedAvg[i], ewma.avg(), 0.1)
+				assert.InDelta(t, tc.expectedVar[i], ewma.varr(), 0.1)
+			}
+		})
+	}
+}