clear
clc
%pca處理
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
fid?=?fopen（‘E:\畢業設計\data\TrainSample.txt‘‘r‘）;
[Acount]=fscanf（fid‘%d?%d‘[5002000]）;
fclose（fid）;

%詞語特征-文檔矩陣?及其?協方差矩陣
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;

%經pca處理后的樣本
global?P;
P?=?A*W_KL‘;
P?=?P‘;
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

%k均值聚類
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
k=500;%聚類后的數目
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）;
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

%