func q6() {
runs := float64(10000)
interval1 := linear.NewInterval(0, 1)
interval2 := linear.NewInterval(0, 1)
sumE1 := float64(0)
sumE2 := float64(0)
sumE := float64(0)
for i := 0; i < int(runs); i++ {
e1 := interval1.RandFloat()
e2 := interval2.RandFloat()
sumE1 += e1
sumE2 += e2
sumE += min(e1, e2)
}
fmt.Printf("e1 = %f, e2 = %f, e = %f\n", sumE1/runs, sumE2/runs, sumE/runs)
}
// NewLinearRegression creates a linear regression object.
// TrainingPoints = 10
// Interval [-1 : 1]
// RandomTargetFunction is true
// Noise = 0
// VectorSize = 3
//
func NewLinearRegression() *LinearRegression {
lr := LinearRegression{
TrainingPoints: 10, // default training points is 10
Interval: linear.NewInterval(-1, 1), // default interval is [-1, 1]
RandomTargetFunction: true, // default RandomTargetFunction is true
Noise: 0, // default noise is 0
VectorSize: 3, // default vector size is 3
}
return &lr
}
func NewBiasAndVariance() *BiasAndVariance {
bav := BiasAndVariance{}
bav.Interval = linear.NewInterval(float64(-1), float64(1))
bav.Runs = 1000
bav.TargetFunction = func(x float64) float64 {
return math.Sin(math.Pi * x)
}
bav.TrainingExampleSize = 2
bav.ThroughOrigin = true
return &bav
}