Finding probability distributions associated with a cross-validated svm using bayesopt

I am finding difficulty in computing the probability of the predictions after training a Support Vector Machine with kfold cross validation and optimizing the hyperparameters using Bayesian optimization.

This is the code I am using

data = [S' U']'; size1 = size(S,1); size2 = size(U,1); theclass = ones((size1+size2),1); theclass(size1+1:end) = -1;

%% Preparing Cross Validation

c = cvpartition((size1+size2),'KFold',100);

%% Optimizing the SVM Classifier

opts = struct('Optimizer','bayesopt','ShowPlots',true,'CVPartition',c,... 'AcquisitionFunctionName','expected-improvement-plus');

svm = fitcsvm(data,theclass,'KernelFunction','rbf',... 'OptimizeHyperparameters','auto','HyperparameterOptimizationOptions',opts)

Any help is appreciated

Answer 1

[Don Mathis]

To get posterior probabilities on a test set using a trained SVM, you can consult this Documentation page:

https://www.mathworks.com/help/stats/fitsvmposterior.html#bt72b34-1