K-medoids聚类 C++代码

以下是K-medoids聚类的C++代码，实现了从数据集中选择k个中心点作为初始值，并迭代更新中心点和簇分配，直到收敛：

#include <iostream>
#include <cmath>
#include <vector>

using namespace std;

double euclidean_distance(const vector<double>& p1, const vector<double>& p2) {
    double sum = 0;
    for (int i = 0; i < p1.size(); i++) {
        sum += pow(p1[i] - p2[i], 2);
    }
    return sqrt(sum);
}

int find_medoid(const vector<vector<double>>& data, const vector<int>& cluster) {
    int medoid = -1;
    double min_cost = numeric_limits<double>::max();
    for (int i = 0; i < data.size(); i++) {
        double cost = 0;
        for (int j = 0; j < data.size(); j++) {
            if (i != j && cluster[j] == cluster[i]) {
                cost += euclidean_distance(data[i], data[j]);
            }
        }
        if (cost < min_cost) {
            min_cost = cost;
            medoid = i;
        }
    }
    return medoid;
}

vector<int> k_medoids(const vector<vector<double>>& data, int k, int max_iter) {
    vector<int> clusters(data.size(), -1);
    vector<int> medoids(k, -1);
    // 初始化中心点
    for (int i = 0; i < k; i++) {
        medoids[i] = rand() % data.size();
        for (int j = 0; j < i; j++) {
            if (medoids[i] == medoids[j]) {
                i--;
                break;
            }
        }
        clusters[medoids[i]] = i;
    }
    for (int iter = 0; iter < max_iter; iter++) {
        // 簇分配
        for (int i = 0; i < data.size(); i++) {
            if (clusters[i] == -1) {
                double min_dist = numeric_limits<double>::max();
                int min_medoid = -1;
                for (int j = 0; j < k; j++) {
                    double dist = euclidean_distance(data[i], data[medoids[j]]);
                    if (dist < min_dist) {
                        min_dist = dist;
                        min_medoid = j;
                    }
                }
                clusters[i] = min_medoid;
            }
        }
        // 更新中心点
        bool changed = false;
        for (int i = 0; i < k; i++) {
            int medoid = find_medoid(data, clusters, i);
            if (medoid != medoids[i]) {
                medoids[i] = medoid;
                clusters[medoid] = i;
                changed = true;
            }
        }
        if (!changed) {
            break;
        }
    }
    return clusters;
}

int main() {
    vector<vector<double>> data = {{2, 6}, {3, 4}, {3, 8}, {4, 7}, {6, 2}, {6, 4}, {7, 3}, {7, 4}, {8, 5}, {7, 6}};
    int k = 2;
    int max_iter = 100;
    vector<int> clusters = k_medoids(data, k, max_iter);
    for (int i = 0; i < clusters.size(); i++) {
        cout << i << ": " << clusters[i] << endl;
    }
    return 0;
}