[R] Time Series Decomposition On Zoo Objects: Errors

Rich Shepard rshepard at appl-ecosys.com
Thu Apr 3 23:54:22 CEST 2014


On Thu, 3 Apr 2014, Roy Mendelssohn wrote:

> How irregular is irregular. kalman filter based methods, such as those in
> KFAS and DLM, can handle missing data, and often "irregular" data can be
> thought of as regular data with missing values, A lot depends on how
> irregular and how big the gaps, to the point where the analysis can be
> calculated but is not very meaningful.

Roy,

   The degree of irregularity varies with the data set. It varies greatly.
Sometimes, monthly samples are missed because the streambed is dry, or
because there are several meters of snow on top of the channel. Sometimes
regulators have permit holders stop analyzing for a chemical constituent,
then there's an agency staff change and that constituent is back on the list
of chemicals to be monitored. Some times fish are present, other times
they're not.

> Don't know if this helps.

   Yes, it does. I'll read up on Kalman filters, on state-space models with
missing data modifications in Shumway & Stoffer (3rd Ed.), and on gamm.

Thanks,

Rich




More information about the R-help mailing list