Hi,

I find the floating search very useful but it would be nice to be able to evaluate another classifier than the default one (1-NN).

best regards,

JM

Hi,

happy to hear that! We will implement other criteria in future releases.

Kind Regards,

Pavel

Hi Pavel,

thanks you!
I have another proposition. Usually the subset of N-1 best features is included in the subset of N best feastures (that’s what I could experience until now). However unless recursively asking the N best features, then the N-1 best, then N-2 etc. it is impossible to have a robust ranking of the the N best features.

It would be nice to fix the number or surviving feature as a vector.For instance sdfeatsel(A,’Crit’,sdsvc([],’C’,[1 10 100],’sigma’,[ 1 3 6 10]),’feat’,1:10) i.e. return the 10,9,8,...best features.

greetings,
JM