Hi. I have been trying to use the Rssa singular spectrum analysis library on a
dataset of 15 minute data at four (cohesive) water temperature stations in an
estuary. The stations have different missing data patterns, though I can find a
period of 2 years where all four are mostly present. There are two major
periodicities (annual and diurnal), and I am using L=N/2. The only calls I am
making are to ssa, plot() to examine wcor and create the groups (total of 6)
and then reconstruct().
The ssa() command with kind="mssa" seems to function to completion. However
each component (e.g. F1, F2) in the reconstruction has the expected dimension
but only one column with non-missing values. The others are NA for every time
point in the column.
The EuStockMarkets mssa example in the documentation works fine for me. When I
pepper it with missing data I get missing data in the reconstruction in those
elements in the data array that were missing in the original data, not entire
columns or any expansion. It seems like my data is causing a big expansion of
missing data. Is there something else I should be doing? For instance should I
be interpolating very small gaps? Why is the first column seemingly always
healthy or is that just a coincidence?
Thanks.
[[alternative HTML version deleted]]
______________________________________________
R-help@r-project.org mailing list -- To UNSUBSCRIBE and more, see
https://stat.ethz.ch/mailman/listinfo/r-help
PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
and provide commented, minimal, self-contained, reproducible code.
