MATLAB中PCA降维程序

PCA降维程序，可以对训练集和测试集同时降维，也可以对单个矩阵进行降维。P为累计贡献率，可以查看降维后保留了百分之多少的有效信息。

第一个对单独矩阵进行降维
% function [p,trainFeatures] = PCA_DimensionReduct(trainFeatures,dimensionReduct)
% %PCA降维
% [totalTrainNum,~] = size(trainFeatures);
% % [totalTestNum,~] = size(testFeatures);
% [~,score,latent,~] = pca(trainFeatures);
% p=cumsum(latent)/sum(latent);%累积贡献值
% trainFeatures = score(1:totalTrainNum,1:dimensionReduct);
% % testFeatures = score(totalTrainNum+1:totalTrainNum+totalTestNum,1:dimensionReduct);
% end

同时对多个矩阵进行降维
function [p,trainFeatures,testFeatures] = PCA_DimensionReduct(trainFeatures,testFeatures,dimensionReduct)
%PCA降维
[totalTrainNum,~] = size(trainFeatures);
[totalTestNum,~] = size(testFeatures);
[~,score,latent,~] = pca([trainFeatures;testFeatures]);
p=cumsum(latent)./sum(latent);%累积贡献值
trainFeatures = score(1:totalTrainNum,1:dimensionReduct);
testFeatures = score(totalTrainNum+1:totalTrainNum+totalTestNum,1:dimensionReduct);
end