I am studing the SSA (Singular Spectrum Analysis) algorithm and after reading some papers about SSA, the ideia was not consolidated.
I found a Matlab SSA algorithm here and I was trying to implement my own SSA, just to get more understanding about it.
My question is just about the SVD (Singular Value Decomposition) step. I would like to know how to substitute the SVD from this code to the svd() implemented by Matlab. I did a comparison of results between the svd() by Matlab and the SVD implemented by this code and the only difference is the part V=(X')*U; that V is different from the V from svd() of Matlab. Why this two SVD are differents ? How can I just use the svd() implemented by Matlab to implement the SSA ?
note: I have an SVD algorithm implemented in other language than Matlab and I want to implement SSA reusing this SVD code.
note 2: X is a time series and the SVD is performed on XX'.