Displaying 1 result from an estimated 1 matches for "modeloptimaic".
2011 Jan 05
0
Nnet and AIC: selection of a parsimonious parameterisation
...axit=maxit,...)
choice[j]=nntmp$choice; nparam[j]=nntmp$nparam;
nparamInit[j]=nntmp$nparaminit; AIC[j]=nntmp$AIC
if (nntmp$AIC<=minAIC) nn=nntmp$nn1
}
#Display the result
tr = cbind(size=size, decay=lambdaj, fit=sqrt(choice/100), nparam=nparam,
nparamInit=nparamInit, AIC=AIC)
modelOptimAIC=order(AIC)[1]
return(list(tr=tr,modelOptimAIC=modelOptimAIC,nn=nn))
}
(nn.res = CVnn.cpus(log10(perf) ~ ., data = cpus3[cpus.samp,]))