What is the difference between fitcsvm and svmtrain?

These two functions aim to do the same thing, but have different options... What is the difference? Why does matlab include both?

Also how does the 'rbf_sigma' paramter of svmtrain compare with the scale paramter of fitcsvm?

Answer 1

[TED MOSBY]

Hi,

Please note that the ‘svmtrain’ function was used in order versions on MATLAB and was removed in R2019b, the function ‘fitcsvm’ is now used place of ‘svmtrain’.

Refer to this MATLAB answer here:

www.mathworks.com/support/search.html/answers/518677-function-svmtrain-not-found.html?fq%5B%5D=asset_type_name:answer&fq%5B%5D=category:matlab/index&page=1

To learn more about ‘fitcsvm’ refer to its MATLAB documentation here:

www.mathworks.com/help/stats/fitcsvm.html?searchHighlight=fitcsvm&s_tid=srchtitle_support_results_1_fitcsvm

To know more about the ‘KernelScale’ parameter refer this link:

www.mathworks.com/help/stats/fitcsvm.html?searchHighlight=fitcsvm&s_tid=srchtitle_support_results_1_fitcsvm#bt9w6j6_sep_shared-KernelScale[UN3] [UN4] [KS5]

MATLAB used to include both functions for backward compatibility. Users who had existing codebases could use ‘svmtrain’ and could continue to use their code without modification. However, for new projects and to take advantage of improved functionality and performance, ‘fitcsvm’ is the recommended choice.

Hope it helps!