SVM with cross-validation

2 vues (au cours des 30 derniers jours)

jason beckell le 8 Mar 2018

0
Lien

Utiliser le lien direct vers cette question

https://fr.mathworks.com/matlabcentral/answers/386992-svm-with-cross-validation

Modifié(e) : jason beckell le 8 Mar 2018

Dear all,

sorry for disturbing you all again, but I have a silly question about SVM with cross-validation. If I write the following code lines:

Mdl = fitclinear(x,y,'ClassNames',[0 1],'KFold',5);
[Y_hat,~] = predict(Mdl,X);

I get an exception due to the fact that Mdl.Trained contains 5 different classification models estimated by means of cross-validation. Hence, I would like to know if there is an automatically way to select the most performing model from the five models reported in Mdl.Trained.

Many thanks for your kind attention! Best,

Fabio.