K-最近邻程序总是报告相同的 class 值

Question

我已经编写了一个 KNN 算法的简短实现，用于根据鸢尾花数据集的简短片段确定样本点 {5.2,3.1} 的 class，但是 class始终报告为 1 (Virginica)。在我的代码中出现问题的地方对我来说并不是很明显。有人可以帮我弄清楚 where/why 发生这种情况吗？

#include <iostream>
#include <math.h>
#include <string>

//Setosa = 0, Virginica = 1, Verscicolor = 2
//[0] and [1] = data point, [2] = class, [3] = distance
double train_data[15][3] = {
{5.3,3.7,0},{5.1,3.8,0},{7.2,3.0,1},
{5.4,3.4,0},{5.1,3.3,0},{5.4,3.9,0},
{7.4,2.8,1},{6.1,2.8,2},{7.3,2.9,1},
{6.0,2.7,2},{5.8,2.8,1},{6.3,2.3,2},
{5.1,2.5,2},{6.3,2.5,2},{5.5,2.4,2}
};

double Distance(double attr1, double attr2, double sAttr1, double sAttr2)
        {
            return sqrt(pow(attr1-sAttr1, 2.0)+pow(attr2-sAttr2, 2.0));
        }

int findMaxIndex(float *classes)
    {
        int maxIndex = 0;
        for (int i = 0; i < 3; i++){
            if (classes[i] > classes[maxIndex])
            maxIndex = i;
        }
        return 0;
    }

 int main(){

    for(int i = 0; i < 15; i++){
                train_data[i][3] = Distance(train_data[i][0],train_data[i][1],5.2,3.1);
    }

    for(int i = 0; i < 15; i++){
                for (int j = i+1; j < 15; j++){

                    if (train_data[i][3] < train_data[j][3]){
                        //swap
                        for(int k = 0; k < 4; k++){
                            double temp = train_data[i][k];
                            train_data[i][k] = train_data[j][k];
                            train_data[j][k] = temp;
                        }
                    }
                }
    }   

    //Based on a value for k determine the class
            int K = 5;
            float *classes = new float[3];
            for (int i =0; i < 3; i++){
                classes[i] = 0;
            }
            for (int i = 0 ; i < K; i++)
            {
                classes[(int)train_data[i][2]-1]++;
            }
            
        int predictedLabel = findMaxIndex(classes)+1;
        std::cout << "Predicted class for point {5.2,3.1} is: " << predictedLabel << std::endl;
        return 0;
}

Answer 1

如果启用警告，您会看到

test.cpp|33 col 32| warning: array subscript 3 is above array bounds of ‘double [3]’ [-Warray-bounds]
||    33 |                 train_data[i][3] = Distance(train_data[i][0],train_data[i][1],5.2,3.1);

数组索引从 0 开始。

稍后，您还减去 (class - 1) 以索引 classes 数组。哎呀。那已经是从零开始的，所以它会变成负数。

考虑避免整个错误来源：

struct {
    double x, y;
    int _class = 0;
    double distance = 0;
} train_data[] = {
    {5.3, 3.7, 0}, {5.1, 3.8, 0}, {7.2, 3.0, 1}, //
    {5.4, 3.4, 0}, {5.1, 3.3, 0}, {5.4, 3.9, 0}, //
    {7.4, 2.8, 1}, {6.1, 2.8, 2}, {7.3, 2.9, 1}, //
    {6.0, 2.7, 2}, {5.8, 2.8, 1}, {6.3, 2.3, 2}, //
    {5.1, 2.5, 2}, {6.3, 2.5, 2}, {5.5, 2.4, 2}  //
};

for(auto& node : train_data) {
    node.distance = Distance(node.x, node.y, 5.2, 3.1);
}

另请注意，std::swap(node1, node2) 将正常工作。或者使节点类型可排序。

更多地道的 C++

这是我的看法

Live On Coliru

#include <array>
#include <iostream>
#include <cmath>
#include <string>
#include <vector>

//Setosa = 0, Virginica = 1, Verscicolor = 2
enum Class { Setosa, Virginica, Verscicolor, NCLASSES };

struct node {
    double x, y;
    Class  _class   = Setosa;
    double distance = 0;

    bool operator<(node const& other) const {
        return distance > other.distance;
    }
};

std::vector<node> train_data{
    {5.3, 3.7, Setosa},      {5.1, 3.8, Setosa},      {7.2, 3.0, Virginica},   //
    {5.4, 3.4, Setosa},      {5.1, 3.3, Setosa},      {5.4, 3.9, Setosa},      //
    {7.4, 2.8, Virginica},   {6.1, 2.8, Verscicolor}, {7.3, 2.9, Virginica},   //
    {6.0, 2.7, Verscicolor}, {5.8, 2.8, Virginica},   {6.3, 2.3, Verscicolor}, //
    {5.1, 2.5, Verscicolor}, {6.3, 2.5, Verscicolor}, {5.5, 2.4, Verscicolor}  //
};

double Distance(double x1, double y1, double x2, double y2) {
    return sqrt(pow(x1 - x2, 2) + pow(y1 - y2, 2));
}

// Based on a value for k determine the class
Class predict(double x, double y, unsigned K) {
    for (auto& node : train_data) {
        node.distance = Distance(node.x, node.y, x, y);
    }
    sort(train_data.begin(), train_data.end());

    // tally buckets:
    std::array<unsigned, NCLASSES> classes = {0, 0, 0};

    for (unsigned i = 0; i < K; i++) {
        classes[train_data[i]._class]++;
    }

    auto index = std::max_element(classes.begin(), classes.end()) //
        - classes.begin();
    return static_cast<Class>(index);
}

int main()
{
    std::cout << "Predicted class for point {5.2,3.1} is: "
              << predict(5.2, 3.1, 5) << "\n";
}

版画

Predicted class for point {5.2,3.1} is: 1

我怀疑你的排序顺序倒置了？翻转排序顺序：

Live On Coliru

Predicted class for point {5.2,3.1} is: 0

K-最近邻程序总是报告相同的 class 值

K-Nearest Neighbors program always reports same class value

c++

classification

machine-learning

knn

更多地道的 C++