What I'm trying to do is dimensionality reduction of a set of m log files containing n time-related signals. They are all physical signals so they are all related to time t.
Consider the dataset descibed below:
-> gridpoints = [t1, ..., tm]
where all the tm vectors are of length T (the temporal vector of each log file)
-> datamatrix =
[f11, ..., fn1]
[f12, ..., fn2]
[..., ..., ...]
[f1m, ..., fnm]
where all the fnm are vectors of length T corrensponding to the m-th observation of the n-th signal.
I'd like to perform FPCA but @vnmabus told me that this feature isn't supported for vector-valued functions yet.
What I'm trying to do is dimensionality reduction of a set of m log files containing n time-related signals. They are all physical signals so they are all related to time t.
Consider the dataset descibed below:
-> gridpoints = [t1, ..., tm] where all the tm vectors are of length T (the temporal vector of each log file)
-> datamatrix = [f11, ..., fn1] [f12, ..., fn2] [..., ..., ...] [f1m, ..., fnm] where all the fnm are vectors of length T corrensponding to the m-th observation of the n-th signal.
I'd like to perform FPCA but @vnmabus told me that this feature isn't supported for vector-valued functions yet.
Here the question I posted on StackOverflow a month ago: https://stackoverflow.com/questions/76012749/how-to-setup-data-with-n-observations-of-m-variables-to-perform-fda-with-scikit