加载电离层data set.

load电离层

Train an SVM classification model using the entire data set. Specify to standardize the data.

MDL= fitcsvm(X,Y,“标准化”,真正的)

MDL= ClassificationSVM ResponseName: 'Y' CategoricalPredictors: [] ClassNames: {'b' 'g'} ScoreTransform: 'none' NumObservations: 351 Alpha: [90×1 double] Bias: -0.1343 KernelParameters: [1×1 struct] Mu: [1×34 double] Sigma: [1×34 double] BoxConstraints: [351×1 double] ConvergenceInfo: [1×1 struct] IsSupportVector: [351×1 logical] Solver: 'SMO'

MDLis a分类vmmodel.

将SVM分类模型保存到文件'SVMIonosphere.mat'。

SaveCompactModel(Mdl,'SVMIonosphere'）；

'SVMIonosphere.mat'出现在您目前的工作目录中。SaveCompactModel通过删除预测不需要的属性（例如训练数据）来减少模型的内存足迹。然后，SaveCompactModel保存一个特征的结构数组MDL在'SVMIonosphere.mat'。

加载structure array in'SVMIonosphere.mat'to the Workspace.

compactmdl = loadcompactmodel（'SVMIonosphere')

CompactMdl = classreg.learning.classif.CompactClassificationSVM ResponseName: 'Y' CategoricalPredictors: [] ClassNames: {'b' 'g'} ScoreTransform: 'none' Alpha: [90×1 double] Bias: -0.1343 KernelParameters: [1×1 struct] Mu: [1×34 double] Sigma: [1×34 double] SupportVectors: [90×34 double] SupportVectorLabels: [90×1 double]

CompactMdlis acompactClassificationvm设备进行预测的模型。

Declare a function that classifies new observations using a trained classification model. This example requires a MATLAB® Coder™ license.

加载Fisher的虹膜数据集。

load渔业

培训由SVM二进制学习者组成的ECOC模型。使用SVM模板，标准化数据，保留30％的培训数据，并指定高斯内核。

t = templatesvm（“标准化”,true,'kernelfunction','gaussian'）；CVMDL= fitcecoc(meas,species,“学习者”，t，'Holdout',0.30);

CVMDLis a分类部门存储等的模型：

• A compact ECOC model, whichfitcecoctrained using the training set.

• 数据分区对象，指定如何fitcecocrandomly split the data into training and holdout sets.

Extract the compact ECOC classification model and the holdout set indices using dot notation. Identify the holdout observations.

compactmdl = cvmdl.Trained {1};cvp = cvmdl.partition;idxtest = test（CVP）;MEATHO = MEAS（IDXTEST，:);

CompactMdlis aCompactClassificationECOCmodel equipped to classify new observations.

Save the compact ECOC classification model to the fileECOCIris.mat。

SaveCompactModel(CompactMdl,'ecociris'）；

Declare a function in your current working folder called分类那：

• Accepts iris-flower measurements commensurate with测量并返回预测标签

• 加载紧凑的ECOC分类模型

• Passes the loaded model and iris-flower measurements to预测

功能label = classifyIrises(X)%#codegen百分比使用ECOC模型对虹膜物种进行分类% CLASSIFYIRISES classifies the iris-flower measurements in X％在文件ecociris.mat中使用紧凑型ECOC模型，然后% returns class labels in label.compactmdl = loadcompactmodel（'ecociris'）；label = predition（compactmdl，x）;end

The%#codegencompilation directive indicates that the MATLAB® code is intended for code generation.

Generate a MEX function from分类。Because C uses static typing,codegenmust determine the properties of all variables in MATLAB® files at compile time. To ensure that the MEX function can use the same input, specify the holdout observations as arguments to the function using the'-args'选项。

codegen分类-args{measHO}

MEX文件classifyiries_mex.mexw64生成您当前的工作目录。文件扩展名取决于您的平台。

Compare labels predicted using预测at the command line,分类， 和classifyIrises_mex。

label1 = predition（compactmdl，measho）;label2 = classifyirise（measho）;label3 = classifyirises_mex（meadho）;comp12 = cellfun（@strcmp，label1，label2）;Comp23 = CellFun（@strcmp，label2，label3）;class12 = sum（comp12）== numel（label1）colle23 = sum（comp23）== numel（label1）

agree12 = logical 1 agree23 = logical 1

标签预测所有三种方式都是相同的。