学习笔记155—机器学习之分类器——Matlab中各种分类器的使用总结,随机森林、支持向量机、K近邻分类器、朴素贝叶斯等

Matlab中常用的分类器有随机森林分类器、支持向量机(SVM)、K近邻分类器、朴素贝叶斯、集成学习方法和鉴别分析分类器等。各分类器的相关Matlab函数使用方法如下:

首先对以下介绍中所用到的一些变量做统一的说明:

train_data——训练样本,矩阵的每一行数据构成一个样本,每列表示一种特征

train_label——训练样本标签,为列向量

test_data——测试样本,矩阵的每一行数据构成一个样本,每列表示一种特征

test_label——测试样本标签,为列向量

①随机森林分类器(Random Forest)

TB=TreeBagger(nTree,train_data,train_label);

predict_label=predict(TB,test_data);

②支持向量机(Support Vector Machine,SVM)

SVMmodel=svmtrain(train_data,train_label);

predict_label=svmclassify(SVMmodel,test_data);

③K近邻分类器(KNN)

KNNmodel=ClassificationKNN.fit(train_data,train_label,\'NumNeighbors\',1);

predict_label=predict(KNNmodel,test_data);

④朴素贝叶斯(Naive Bayes)

Bayesmodel=NaiveBayes.fit(train_data,train_label);

predict_label=predict(Bayesmodel,test_data);

⑤集成学习方法(Ensembles for Boosting)

Bmodel=fitensemble(train_data,train_label,\'AdaBoostM1\',100,\'tree\',\'type\',\'classification\');

predict_label=predict(Bmodel,test_data);

⑥鉴别分析分类器(Discriminant Analysis Classifier)

DACmodel=ClassificationDiscriminant.fit(train_data,train_label);

predict_label=predict(DACmodel,test_data);

具体使用如下:(练习数据下载地址如下http://en.wikipedia.org/wiki/Iris_flower_data_set,简单介绍一下该数据集:有一批花可以分为3个品种,不同品种的花的花萼长度、花萼宽度、花瓣长度、花瓣宽度会有差异,根据这些特征实现品种分类)

%% 随机森林分类器(Random Forest)

nTree=10;

B=TreeBagger(nTree,train_data,train_label,\'Method\', \'classification\');

predictl=predict(B,test_data);

predict_label=str2num(cell2mat(predictl));

Forest_accuracy=length(find(predict_label == test_label))/length(test_label)*100;

%% 支持向量机

% SVMStruct = svmtrain(train_data, train_label);

% predictl=svmclassify(SVMStruct,test_data);

% predict_label=str2num(cell2mat(predictl));

% SVM_accuracy=length(find(predict_label == test_label))/length(test_label)*100;

%% K近邻分类器(KNN)

% mdl = ClassificationKNN.fit(train_data,train_label,\'NumNeighbors\',1);

% predict_label=predict(mdl, test_data);

% KNN_accuracy=length(find(predict_label == test_label))/length(test_label)*100

%% 朴素贝叶斯 (Naive Bayes)

% nb = NaiveBayes.fit(train_data, train_label);

% predict_label=predict(nb, test_data);

% Bayes_accuracy=length(find(predict_label == test_label))/length(test_label)*100;

%% 集成学习方法(Ensembles for Boosting, Bagging, or Random Subspace)

% ens = fitensemble(train_data,train_label,\'AdaBoostM1\' ,100,\'tree\',\'type\',\'classification\');

% predictl=predict(ens,test_data);

% predict_label=str2num(cell2mat(predictl));

% EB_accuracy=length(find(predict_label == test_label))/length(test_label)*100;

%% 鉴别分析分类器(discriminant analysis classifier)

% obj = ClassificationDiscriminant.fit(train_data, train_label);

% predictl=predict(obj,test_data);

% predict_label=str2num(cell2mat(predictl));

% DAC_accuracy=length(find(predict_label == test_label))/length(test_label)*100;

%% 练习

% meas=[0 0;2 0;2 2;0 2;4 4;6 4;6 6;4 6];

% [N n]=size(meas);

% species={\'1\';\'1\';\'1\';\'1\';\'-1\';\'-1\';\'-1\';\'-1\'};

% ObjBayes=NaiveBayes.fit(meas,species);

% x=[3 3;5 5];

% result=ObjBayes.predict(x);

参考链接:https://blog.csdn.net/jisuanjiguoba/java/article/details/80004568