Dealing with Unwanted Observations

[alexecon]

I have a time-series variable ("VOLUME") that contains zeros. I would like to drop/delete/ignore observations that correspond to these zeros in all variables in the dataset. The following

Code: Select all

SMPL(SERIES=VOLUME) 2016:01:01//1 2017:02:10//23

does not subsequently allow me to, e.g., estimate a GARCH model and

Code: Select all

GARCH(P=1,Q=1)

gives instead

GARCH Cannot Be Used with Gaps/Missing Values

Would it be better to get rid of the unwanted observations from the data at the source or can RATS do that? Thank you.

[TomDoan]

Read the part of the next page beginning with "Another option"

https://estima.com/docs/RATS%209%20Intr ... f#page=116

Your SMPL option would be based upon the volume being zero, rather than the %VALID (which tests for missing values), but the idea is the same.