Saving nested for loop output in a single column

Hello everyone,

I am estimating time-series models with different AR, MA, and lag components - am using nested for loops. My problem is simple but am new to MATLAB environment. Issue: I am scoring different information criteria (such as AIC) within the loop structure. I would like to put the outputs of information criteria into a new column, one for each of the information criteria (see below pl).

Below is what I have tried. Rather than putting the outputs for a specific information criteria in one column, it creates an array with x*y*z form. I am sure I am missing something very obvious in the code below, especially in the last lines. Any help will be truly appreciated!

pmax = 3;       %Max order of AR   
dmax = 1;       %Max order of difference
qmax = 3;     % Max order of MA                

%Pre-allocation
a = pmax + 1;
b = dmax + 1;
c = qmax + 1;
aa = combnk(a,1);
bb = combnk(b,1);
cc = combnk(c,1);
combinations = (aa*bb*cc)+1;

aiccol = zeros(combinations,1);
caiccol = zeros(combinations,1);
icompcol = zeros(combinations,1);

t = 499;
vt = Data(:,4);
n=length(vt);
for k = 0:pmax
    for j = 0:dmax
    for i = 0:qmax
Md1 = arima(k,j,i);
[EstMd1,EstParamcov,logL]= estimate(Md1,vt);
[res,~,logL] = infer(EstMd1,vt);
sig2 = (vt'*vt)/t;
s=rank(EstParamcov);
tr=trace(EstParamcov);
d=det(EstParamcov);
AIC=n*log(2*pi)+n*log(sig2)+n+2*(k+i+1)
CAIC=n*log(2*pi)+n*log(sig2)+n+2*(k+i+1)*(log(n)+1)
ICOMP=n*log(2*pi)+n*log(sig2)+n+(k+i+1)+s*log(tr/s)-2*log(abs(d))
aiccol(k+1,j+1,i+1) = AIC;                   %?????             *Something wrong here*
caiccol(k+1,j+1,i+1) = CAIC;                 %?????             *Something wrong here*
icompcol(k+1,j+1,i+1) = ICOMP;               %?????             *Something wrong here*
    end
    end
end

Answer 1

[KSSV]

Try this..

pmax = 3;       %Max order of AR
dmax = 1;       %Max order of difference
qmax = 3;     % Max order of MA
%Pre-allocation
a = pmax + 1;
b = dmax + 1;
c = qmax + 1;
aa = combnk(a,1);
bb = combnk(b,1);
cc = combnk(c,1);
combinations = (aa*bb*cc)+1;
% aiccol = zeros(combinations,1);
% caiccol = zeros(combinations,1);
% icompcol = zeros(combinations,1);
aiccol = zeros(pmax+1,dmax+1,qmax+1) ;
caiccol = zeros(pmax+1,dmax+1,qmax+1) ;
icompcol = zeros(pmax+1,dmax+1,qmax+1) ;
t = 499;
vt = Data(:,4);
n=length(vt);
for k = 0:pmax
    for j = 0:dmax
        for i = 0:qmax
            Md1 = arima(k,j,i);
            [EstMd1,EstParamcov,logL]= estimate(Md1,vt);
            [res,~,logL] = infer(EstMd1,vt);
            sig2 = (vt'*vt)/t;
            s=rank(EstParamcov);
            tr=trace(EstParamcov);
            d=det(EstParamcov);
            AIC=n*log(2*pi)+n*log(sig2)+n+2*(k+i+1)
            CAIC=n*log(2*pi)+n*log(sig2)+n+2*(k+i+1)*(log(n)+1)
            ICOMP=n*log(2*pi)+n*log(sig2)+n+(k+i+1)+s*log(tr/s)-2*log(abs(d))
            aiccol(k+1,j+1,i+1) = AIC;                   %?????             *Something wrong here*
            caiccol(k+1,j+1,i+1) = CAIC;                 %?????             *Something wrong here*
            icompcol(k+1,j+1,i+1) = ICOMP;               %?????             *Something wrong here*
        end
    end
end