i am trying to use cross validation in order to determine the optimum number of hidden units for neural network. Am getting an error which i am not able to decipher.

Error using crossval>evalFun (line 480) The function '@(Xtrain,Ytrain,Xtest)model_finder(i,Xtrain,Ytrain,Xtest)' generated the following error: Invalid types for comparison.

Error in crossval>getLossVal (line 517) funResult = evalFun(funorStr,arg(1:end-1));

Error in crossval (line 416) [funResult,outarg] = getLossVal(i, nData, cvp, data, predfun);

Error in nnrealmain (line 7) mcr=crossval('mcr',x,y,'predfun',hid_find,'partition',c);

This is the main code i typed for cross validation.

load('permanentpcadata.mat');

mcrs=[]; y=[ones(18,1);2*ones(13,1);3*ones(18,1);4*ones(16,1);5*ones(21,1);6*ones(9,1)];

for i=6:20

hid_find=@(Xtrain,Ytrain,Xtest)model_finder(hiiden_units,Xtrain,Ytrain,Xtest); c=cvpartition(y,'k',10); mcr=crossval('mcr',x,y,'predfun',hid_find,'partition',c);

mcrs=[mcrs mcr]; end

save('crossvalop.mat','mcrs');

index=6:20; plot(index,mcrs);

could you suggest where i have gone wrong in the implementation of cross validation?

[Greg Heath]

My only suggestion is to search both the NEWSGROUP and ANSWERS using the search word 'crossval'