MATRIX ARRANGEMENT PROBLEM MATLAB
2 Ansichten (letzte 30 Tage)
Ältere Kommentare anzeigen
Hello,
I am working with panel structural breaks, where , and X1, X2, X3 (d=3).
Now I have three breaks, B=3, while the author uses the code with B=1 and GB, GA (for me it will be GI, GII, GIII, GIV)
I am stuck at T_frac, Y_frac, X_frac somehow the index becomes more than 3.
I am permission to use the code of the author, any help will be appreciated.
Xvec = [x1, x2, x3];
Yvec = y;
N=114;
TT=48;
xlag = x(:,1:TT-1,:);
ylag = y(:,2:TT);
T = TT-1;
Xmat = xlag;
ymat = ylag;
p = size(Xmat,3);
S = 30;
SSR = zeros(1,(T-2));
SSR(1) = inf;
ymatfd = diff(ymat,1,2);
Xmatfd = diff(Xmat,1,2);
for j = 1:B
for bk = 2:(T-2)
regime = [0,bk,T-1]; % Get the bounds of potential subperiods
for aa = 1:(j+1)
T_frac = length((regime(aa)+1):(regime(aa+1)));
y_frac = reshape(ymatfd(:,(regime(aa)+1):(regime(aa+1)))',N*T_frac,1);
for l = 1:p
X_frac(:,l) = reshape(Xmatfd(:,(regime(aa)+1):(regime(aa+1)),l)',N*T_frac,1);
end
time_frac = repmat(1:T_frac,1,N)';
code_frac = kron((1:N)',ones(T_frac,1));
idxdata_frac = dataset(code_frac, time_frac, y_frac, X_frac);
idxdata_frac.Properties.VarNames = {'N' 'T' 'y' 'X'};
if aa == 1
[~,~,ssr,] = est_group(idxdata_frac,GB);
elseif aa == 2
[~,~,ssr,] = est_group(idxdata_frac,GA);
end
SSR(bk) = SSR(bk) + ssr;
clear X_frac y_frac idxdata_frac time_frac code_frac
end
end
[resQ,k] = min(SSR); % Compare SSR to get a breakdate
end
2 Kommentare
Jan
am 14 Jun. 2022
"Now I have three breaks, B=3, while the author uses the code with B=1 and GB, GA (for me it will be GI, GII, GIII, GIV)" - remember, that the readers do not know, what you are talking off. Who is the autheor? What us GB and GA, GI, GII, GIII and GIV?
What does this mean: "I am stuck at T_frac, Y_frac, X_frac somehow the index becomes more than 3."?
Antworten (1)
Cris LaPierre
am 14 Jun. 2022
Not that your loop counter is j, but your indexing uses j+1, and then again use loop counter aa, but index using aa+1. Therefore, when B=3, j+1=4 and aa+1=5.
for j = 1:B
...
for aa = 1:(j+1)
T_frac = length((regime(aa)+1):(regime(aa+1)));
% ^^^^
end
end
12 Kommentare
Cris LaPierre
am 16 Jun. 2022
Bearbeitet: Cris LaPierre
am 16 Jun. 2022
I didn't mention a fix. I just pointed out the code that was causing the problem.
The fix is to make regime longer. It needs to be a 1x(B+2) vector to not error. Again, it is currently 1x3, so when B=1, the code works. When B=3, it needs to be 1x5.
I don't know what values to add, though. That is something where you would need to apply your knowledge in this space to corectly adjust the code.
Siehe auch
Kategorien
Mehr zu Logical finden Sie in Help Center und File Exchange
Community Treasure Hunt
Find the treasures in MATLAB Central and discover how the community can help you!
Start Hunting!