K-Means clustering & calculating some stats for nutrition data

main.go

package main

import (
	"fmt"
	"math"

	"github.com/montanaflynn/stats"
	"github.com/muesli/clusters"
	"github.com/muesli/kmeans"
	"gonum.org/v1/gonum/stat"
)

func main() {

	// data in the following format
	// [0]   cosine similarity distance of an item represented by these values
	// [1]   number of calories for this sample
	// [2..] nutrient amounts in grams (carbs, protein etc..)
	bbqChicken := [][]float64{
		{0.30936217308044434, 339, 41.73, 11.28, 1.2, 15.795},
		{0.31012654304504395, 396, 50.4, 14.4, 2.4, 16.8},
		{0.31477395804083474, 357, 49.665, 8.94, 1.95, 16.89},
		{0.32190563213262624, 357, 49.665, 8.94, 1.95, 16.89},
		{0.3601418137550354, 396, 50.4, 14.4, 2.4, 16.8},
		{0.3631199598312378, 340.5, 40.95, 13.05, -1, 14.1},
		{0.3631199598312378, 375, 48, 11.1, -1, 19.5},
		{0.36419129371643066, 369, 53.55, 10.35, -1, 15.45},
		{0.47245991230010986, 620, 19, 33, 0, 57},
		{0.47404394966494867, 365, 8, 28, 1, 19},
		{0.3238506317138672, 313.5, 35.60999999999999, 11.865, 1.05, 17.265},
		{0.32428465259923467, 301.5, 36.69, 9.705, 1.05, 17.265},
		{0.365716814994812, 366, 49.65, 10.95, 3, 16.05},
		{0.3693286588587954, 333, 46.5, 8.4, 3.75, 16.5},
		{0.2780816389273607, 310.5, 42, 9, -1, 13.8},
		{0.296245813369751, 294, 30.6, 10.95, 2.7, 18.3},
		{0.37020546197891235, 349.5, 44.7, 10.5, -1, 17.4},
		{0.41507232189178467, 538, 33, 28, 2, 37},
		{0.4200516939163208, 998, 37, 64, 0, 65},
		{0.43077021837234497, 561, 63, 21, 3, 29},
		{0.432941112939052, 422, 62, 11, 3, 18},
		{0.4425056576728821, 877, 94, 34, 11, 50},
		{0.45972079038619995, 263, 11, 14, 1, 23},
		{0.4625535011291504, 457, 54, 11, 6, 35},
		{0.4642601449567121, 292, -1, 6, 1, 35},
		{0.3245284356036313, 313.5, 35.60999999999999, 11.865, 1.05, 17.265},
		{0.3247811794281006, 301.5, 36.69, 9.705, 1.05, 17.265},
		{0.3260763883590698, 301.5, 36.69, 9.705, 1.05, 17.265},
		{0.32911187410354614, 313.5, 35.60999999999999, 11.865, 1.05, 17.265},
		{0.3515464663505554, 366, 41.7, 12.45, 4.5, 18.9},
	}

	// similar sample, sampling nutrients of a chick-fil-a chicken sandwich
	// chickFila := [][]float64{
	// 	{0.17526186557184698, 373.5, 31.335, 16.785, 2.1, 24.42},
	// 	{0.31898687240442225, 421.5, 41.22, 20.19, 2.55, 18.255},
	// 	{0.32166337966918945, 342, 15.585, 16.8, 1.35, 32.1},
	// 	{0.33655810356140137, 436.5, 36.96, 23.1, 2.25, 19.86},
	// 	{0.33884588844502517, 339, 29.1, 9.27, 1.05, 32.25},
	// 	{0.3511131808493486, 364.5, 26.28, 13.665, 0.9, 32.1},
	// 	{0.3525458574295044, 450, 36.585, 25.109999999999996, 3, 19.08},
	// 	{0.3615299272993182, 421.5, 41.22, 20.19, 2.55, 18.255},
	// 	{0.3642024614931608, 379.5, 25.92, 15.72, 1.8, 31.305},
	// 	{0.38329021284849063, 375, 31.335, 16.785, 2.1, 24.42},
	// 	{0.40265870094299316, 253.5, 35.25, 6.9, -1, 15.45},
	// 	{0.41297308129232535, 330, 48, 7.05, -1, 15},
	// 	{0.4183231245471899, 357, 34.2, 12.6, -1, 26.85},
	// 	{0.4253319501876831, 312, 27.9, 15.75, -1, 13.65},
	// 	{0.4301462769508362, 333, 37.35, 10.545, 3.24, 19.5},
	// 	{0.4358830451965332, 286.5, 26.85, 11.4, -1, 16.5},
	// 	{0.43745875358581543, 285, 29.25, 10.95, 3.45, 15.6},
	// 	{0.44068199396133423, 318, 21, 12.6, -1, 36},
	// 	{0.44141894578933716, 253.5, 36, 4.65, 5.7, 13.95},
	// 	{0.44735704205504156, 214.29000000000005, 0, 10.71, -1, 29.46},
	// 	{0.47878752585612294, 3028, 37, 324, 4, 4},
	// 	{0.47987449169158936, 688, 76, 27, 6, 35},
	// 	{0.4989275782980791, 1478, 165, 61, 9, 64},
	// 	{0.5153046250343323, 442, 42, 14, 2, 35},
	// }

	rawData := bbqChicken

	var d clusters.Observations

	for _, v := range rawData {
		coordinates := clusters.Coordinates(v)
		d = append(d, coordinates)
	}

	k := int(math.Sqrt(float64(len(rawData))))
	fmt.Println("Using k:", k)

	km := kmeans.New()
	clusters, err := km.Partition(d, k)

	if err != nil {
		fmt.Println(err)
		return
	}

	for _, c := range clusters {
		// fmt.Printf("Centered at x: %.2f y: %.2f\n", c.Center[0], c.Center[1])
		fmt.Printf("Matching data points: %+v\n\n", c.Observations)

		var dist []float64
		var cals []float64

		for _, ob := range c.Observations {
			dist = append(dist, ob.Coordinates().Coordinates()[0])
			cals = append(cals, ob.Coordinates().Coordinates()[1])
		}

		md, mdstd := stat.MeanStdDev(dist, nil)

		weights := make([]float64, len(dist))
		for i := range dist {
			weights[i] = 1 - dist[i]
		}

		medianDist, err := stats.Median(dist)

		if err != nil {
			panic(err)
		}

		iqr, err := stats.InterQuartileRange(dist)

		if err != nil {
			panic(err)
		}

		lowest, err := stats.Min(dist)

		if err != nil {
			panic(err)
		}

		calMean, calStd := stat.MeanStdDev(cals, weights)

		fmt.Printf("Distance Min: %.2f Mean: %.2f Std: %.2f Median: %.2f IQR: %.2f\n", lowest, md, mdstd, medianDist, iqr)
		fmt.Printf("Mean cals: %.2f Stdev: %.3f\n\n", calMean, calStd)
	}
}