pca源码matlab

clear
clc
%pca处理
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
fid = fopen（‘E:\毕业设计\data\TrainSample.txt‘‘r‘）;
[Acount]=fscanf（fid‘%d %d‘[5002000]）;
fclose（fid）;

%词语特征-文档矩阵 及其 协方差矩阵
A = A‘;
R_A = A‘ * A;

%求最大的特征值
d = eig（R_A）;
[V D] = eig（R_A）;
clear R_A;
s = sum（d）;
for i=1:500
    if （sum（d（500-i:500））/s > 0.85）
        break;
    end
end
clear s;
global W_KL;
W_KL = V（:500-i:500）‘;
clear V D;

%经pca处理后的样本
global P;
P = A*W_KL‘;
P = P‘;
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

%k均值聚类
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
k=500;%聚类后的数目
p_cluster（:1:k） = cluster（P（:1:1000） k i+1 1000）;
p_cluster（:k+1:2*k） = cluster（P（:1001:2000） k i+1 1000）;
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

%