Looping a regression for 30 different observations 10 times

Suppose I have a total of 300 observations.

Now I would like to perform a regression over the observation 1:30 31:60 etc.

And I am intersted in the error of the regression

X = [ones(size(x)) x1 x2 x3];

[b,bint,r,rint,stats] = regress(y,X)

How do I write a for loop such that I get 300/30 = 10 values of the error of the regression?

Answer 1

[Star Strider]

One approach —

xa = randn(300,3);
ya = randn(300,1);
forstep = 30;
for k = 1:forstep:size(xa,1)-forstep
    idxrng = k:k+30;
    kidx = floor(k/forstep+1);
    x1 = xa(idxrng,1);
    x2 = xa(idxrng,2);
    x3 = xa(idxrng,3);
    y = ya(idxrng);
    X = [ones(size(x1)) x1 x2 x3];
    [b{kidx},bint{kidx},r{kidx},rint{kidx},stats{kidx}] = regress(y,X);
end
b

b = 1×9 cell array
    {4×1 double}    {4×1 double}    {4×1 double}    {4×1 double}    {4×1 double}    {4×1 double}    {4×1 double}    {4×1 double}    {4×1 double}

bint

bint = 1×9 cell array
    {4×2 double}    {4×2 double}    {4×2 double}    {4×2 double}    {4×2 double}    {4×2 double}    {4×2 double}    {4×2 double}    {4×2 double}

r

r = 1×9 cell array
    {31×1 double}    {31×1 double}    {31×1 double}    {31×1 double}    {31×1 double}    {31×1 double}    {31×1 double}    {31×1 double}    {31×1 double}

... and so for the rest.

.