// ++++++++++++++++++++++++++++++++++++++++
// 《零基础Go语言算法实战》源码
// ++++++++++++++++++++++++++++++++++++++++
// Author:廖显东（ShirDon）
// Blog:https://www.shirdon.com/
// Gitee:https://gitee.com/shirdonl/goAlgorithms.git
// Buy link :https://item.jd.com/14101229.html
// ++++++++++++++++++++++++++++++++++++++++

package main

import (
	"fmt"
	"math"
)

// 表示具有特征和类标签的数据点
type DataPoint struct {
	features []float64
	label    string
}

// 计算两个数据点之间的欧氏距离
func euclideanDistance(p1, p2 DataPoint) float64 {
	var sumSquares float64
	for i := range p1.features {
		diff := p1.features[i] - p2.features[i]
		sumSquares += diff * diff
	}
	return math.Sqrt(sumSquares)
}

// 根据 k 最近邻点对新数据点进行分类
func kNNClassify(k int, trainingSet []DataPoint, newPoint DataPoint) string {
	// 计算新点与训练集中所有点之间的距离
	distances := make([]float64, len(trainingSet))
	for i, point := range trainingSet {
		distances[i] = euclideanDistance(point, newPoint)
	}

	// 找到 k 最近邻点的索引
	indices := make([]int, k)
	for i := range indices {
		minIndex := 0
		for j := range distances {
			if distances[j] < distances[minIndex] {
				minIndex = j
			}
		}
		indices[i] = minIndex
		distances[minIndex] = math.MaxFloat64
	}

	// 计算 k 最近邻点的类标签
	counts := make(map[string]int)
	for _, index := range indices {
		counts[trainingSet[index].label]++
	}

	// 确定 k 最近邻点中的多数类标签
	var (
		maxCount int
		maxLabel string
	)
	for label, count := range counts {
		if count > maxCount {
			maxCount = count
			maxLabel = label
		}
	}

	return maxLabel
}

func main() {
	// 创建具有特征和类标签的数据点训练集
	trainingSet := []DataPoint{
		{[]float64{2.0, 4.0}, "A"},
		{[]float64{4.0, 2.0}, "A"},
		{[]float64{4.0, 4.0}, "B"},
		{[]float64{4.0, 6.0}, "B"},
		{[]float64{6.0, 4.0}, "B"},
	}

	// 创建一个新的数据点进行分类
	newPoint := DataPoint{[]float64{6.0, 6.0}, ""}

	// 使用 KNN 算法对新数据点进行分类
	k := 3
	label := kNNClassify(k, trainingSet, newPoint)

	fmt.Printf("新点属于类：%s\n", label)
}

//$ go run interview10-15.go
//新点属于类：B