n维 Fisher判别 c++实现

要知道具体的原理，可以看其他博客的作品。

代码如下：

#include<iostream>
#include<Eigen\dense>
#include<vector>
using namespace std;
using namespace Eigen;

int Fisher::fisher(vector<VectorXd> x, vector<VectorXd> y, VectorXd test_vector)
//返回1，代表属于第一类；返回2，代表属于第二类。
{
	if (x.size() < 0 || y.size() < 0 || x.size() != y.size())
	{
		cout << "两个类的样本向量个数有误。" << endl;
		return 0;
	}
	if (x[0].size() < 1 || y[0].size() < 1 || x[0].size() != y[0].size())
	{
		cout << "两个类的样本向量维度有误。" << endl;
		return 0;
	}
	if (test_vector.size() != x[0].size())
	{
		cout << "给定的向量有误。" << endl;
		return 0;
	}
	int n = x[0].size();
	VectorXd meanx(n), meany(n);
	for (int i = 0; i < n; i++)
	{
		meanx(i) = 0;
		meany(i) = 0;
	}

	//对每个维度分别求和
	for (int i = 0; i < x.size(); i++)
	{
		meanx += x[i] ;
		meany += y[i];
	}
	
	//分别求平均
	double c = 1.0 / x.size();
	//cout << meanx << endl;
	meanx = meanx * c;
	meany = meany * c;
	//cout << meanx << endl;

	//求类内离散矩阵
	MatrixXd Swx, Swy, Sw;
	Swx.setZero(n, n);
	Swy.setZero(n, n);
	for (int i = 0; i < x.size(); i++)
	{
		Swx += (x[i] - meanx) * (x[i] - meanx).transpose();
		Swy += (y[i] - meany) * (y[i] - meany).transpose();
	}

	//求类内总离散矩阵
	Sw = Swx + Swy;
	cout << "类内总离散矩阵：" << endl << Sw << endl;

	//实现投影
	MatrixXd w, w0;
	w = Sw.inverse() * (meany - meanx);
	w0 = (w.transpose() * meanx + w.transpose() * meany) / 2;
	cout << "最佳投影方向：" << endl << w << endl << "阈值为： " << w0 << endl;
	MatrixXd sample;
	double w0_num = w0(0, 0);
	sample = w.transpose() * test_vector;
	double sample_num = sample(0, 0);
	if (sample_num < w0_num)
		cout << "给定的向量属于第一类。" << endl;
	else
		cout << "给定的向量属于第二类。" << endl;
	return 1;
}

 运行结果如下(基于我自己的数据）: