Rolling Window PCA- Save Explained factors for every period

I need to do rolling window PCA and I have a rolling window size=20. But I want to save all explained factors for [coeff,scores,explained]=pcacov(X) function in for loop.

What would be the most simple solution?

Thank you very much.

Answer 1

[the cyclist]

One way;

numberOfVars = 100; % Set to however big your data are
numberOfWindows = 10;

explainedArray = zeros(numberOfVars,numberOfRuns);

for nw = 1:numberOfWindows
  [coeff,scores,explained]=pcacov(X);
  explainedArray(:,nw) = explained;
end

[meral serçe]

Thank you very much cyclist,

But I need explained factors for every rolling window period and I want to save them in a variable in a workspace.

So for example; for every period of window size=100; I need explained factors and then for other 100 observations I need other explained factors and store them in a workspace.

[the cyclist]

I assumed you only needed help with the storage in the for loop, since that is all you asked about.

Just modify this line

[coeff,scores,explained]=pcacov(X);

so that you have the rolling window within X each time. Maybe something like

X = biggerX(<define window in here>,:)

Do you need help in how to define a rolling window?

[meral serçe]

Do you have a chance to define a rolling window, and write a code for window size=20.

Thank you,

[the cyclist]

Insert a line like this in the loop:

rollingWindowX = fullX(nw:nw+20,:)

and do the PCA on rollingWindowX.