Displaying 1 result from an estimated 1 matches for "aic_k".
Did you mean:
aic_
2008 May 23
0
Est. Component Size with AIC/BIC under Gamma Distribution
...9.7, 12.8)
# initialize
start_ <- 300
error_ <- 0.001
thetastart <- 1
k <- 5
# compute AIC/BIC for k component
for (nofc in 1:k {
cat("k = ", nofc, "\n")
maxlogl <- -(mlogl_process(vsamples,error_,thetastart))
bic_k <- find_bic(maxlogl,vsamples,nofc)
aic_k <- find_aic(maxlogl,vsamples,nofc)
cat("BIC = ", bic_k, " - AIC = ", aic_k, "- MLL= ", maxlogl,"\n")
}
__ END__
We finally expect to choose K with the smallest AIC/BIC
value.
However, the problem I have is that the AIC/BIC value is always
on the incre...