一個(gè)SVM分類的測(cè)試代碼模板

1、精選優(yōu)質(zhì)文檔-傾情為你奉上SVM分類代碼模板，在使用的時(shí)候只需要將數(shù)據(jù)集導(dǎo)入，在修改一點(diǎn)點(diǎn)參數(shù)就可以進(jìn)行分類效果及其可視化顯示。Contents原始數(shù)據(jù)可視化歸一化預(yù)處理feature selection參數(shù)c和g尋優(yōu)選擇分類預(yù)測(cè) %SVM分類測(cè)試程序代碼模板 %清空變量空間 clc;clear;close all format compact %程序代碼計(jì)時(shí)開始 tic; %數(shù)據(jù)導(dǎo)入 X0=load(C:Documents and SettingsAdministrator桌面訓(xùn)練數(shù)據(jù)ecolitrain.txt); Y0=load(C:Documents and SettingsAdmin

2、istrator桌面訓(xùn)練數(shù)據(jù)ecolitest.txt); trainset=X0(:,2:end); trainset_label=X0(:,1); testset=Y0(:,2:end); testset_label=Y0(:,1);原始數(shù)據(jù)可視化 figure; boxplot(trainset,orientation,horizontal); grid on; title(Visualization for original data); %原始樣本集各維度可視化 featuredisplay(trainset,trainset_label,testset,testset_label,

3、1); %計(jì)數(shù)樣本類別數(shù) labelnum=countlabelnum(testset_label);歸一化預(yù)處理 trainset,testset = scaleForSVM(trainset,testset,0,1); %樣本集歸一化后各維度可視化 featuredisplay(trainset,trainset_label,testset,testset_label,1); %降維處理 %1.pca處理 trainset,testset = pcaForSVM(trainset,testset); % trainset,testset = pcaForSVMtxb(trainset,te

4、stset); % %2.ica處理 % trainset,testset = fasticaForSVM(trainset,testset); % %3.dct處理 % trainset,testset = DCTforSVM(trainset,testset); feature selectionusing GA,.,etc.參數(shù)c和g尋優(yōu)選擇 bestacc,bestc,bestg = SVMcgForClass(trainset_label,trainset,-8,8,-8,8,3,0.5,0.5,0.5) cmd = -c ,num2str(bestc), -g ,num2str(b

5、estg);bestacc = 87.7676 bestc = 2 bestg = 4分類預(yù)測(cè) model = svmtrain(trainset_label, trainset,cmd); ptrain_label, train_acc = svmpredict(trainset_label, trainset, model); predict_label, test_acc,dd = svmpredict(testset_label, testset, model);Accuracy = 89.6024% (293/327) (classification) Accuracy = 91.1111% (41/45) (classification) %ROC曲線 plotSVMroc(testset_label,predict_label,labelnum); %分類效果可視化、正確率 rightpercent,testnum,errornum,errorindex=classmeasures(testset_label,predict_label,bestc,bestg);Warning: Ignoring extra legend entries. %程序運(yùn)行