Predicting the Missing Humidity Values

Linear Regression.md

Predicting the Missing Humidity Values

predict-missing-humidity.go

package main

import (
	"fmt"
	"time"

	"github.com/sjwhitworth/golearn/base"
	"github.com/sjwhitworth/golearn/linear_models"
)

func mustParseDT(dtStr string) int64 {
	layout := "2006-01-02 15:04"
	str := dtStr
	t, err := time.Parse(layout, str)
	if err != nil {
		panic(err)
	}
	return t.Unix()
}

func abs(n int64) int64 {
	y := n >> 63
	return (n ^ y) - y
}

func checkerr(err error) {
	if err != nil {
		panic(err)
	}
}

func predictMissingHumidity(startDate string, endDate string, knownTimestamps []string, humidity []float64, timestamps []string) []float64 {
	lm := linear_models.NewLinearRegression()
	grid := base.NewDenseInstances()
	attrs := []base.Attribute{
		base.NewFloatAttribute("Temperature"),
		base.NewFloatAttribute("Timestamp"),
	}
	specs := make([]base.AttributeSpec, len(attrs))
	// Allocate the Instances to return
	for i, a := range attrs {
		spec := grid.AddAttribute(a)
		specs[i] = spec
	}
	checkerr(grid.Extend(len(knownTimestamps)))

	for i, item := range humidity {
		grid.Set(specs[0], i, base.PackFloatToBytes(item))
	}
	for i, item := range knownTimestamps {
		ts := abs(time.Unix(0, 0).Unix() - mustParseDT(item))
		grid.Set(specs[1], i, base.PackFloatToBytes(float64(ts)))
	}

	checkerr(grid.AddClassAttribute(attrs[0]))
	checkerr(lm.Fit(grid))

	gridZ := base.NewStructuralCopy(grid)
	attrsZ := gridZ.AllAttributes()
	specsZ := base.ResolveAttributes(gridZ, attrsZ)
	checkerr(gridZ.Extend(len(timestamps)))
	for i, item := range timestamps {
		ts := abs(time.Unix(0, 0).Unix() - mustParseDT(item))
		//gridZ.Set(specsZ[0], i, base.PackFloatToBytes(0.0))
		gridZ.Set(specsZ[1], i, base.PackFloatToBytes(float64(ts)))
	}

	predicted, err := lm.Predict(gridZ)
	checkerr(err)

	result := make([]float64, len(timestamps))
	for i := range result {
		result[i] = base.UnpackBytesToFloat(predicted.Get(specsZ[0], i))
	}

	return result
}

func main() {
	fmt.Println(predictMissingHumidity("2013-01-01", "2013-01-01", []string{
		"2013-01-01 07:00",
		"2013-01-01 08:00",
		"2013-01-01 09:00",
		"2013-01-01 10:00",
		"2013-01-01 11:00",
		"2013-01-01 12:00",
	}, []float64{
		10.0,
		11.1,
		13.2,
		14.8,
		15.6,
		16.7,
	}, []string{
		"2013-01-01 13:00",
		"2013-01-01 14:00",
	}))
}