Observation IDs within a sample

[jthodge]

Is there a way to create a series for an unstructured workfile that shows the relative observation IDs within a sample?

The @OBSNUM and @OBSID functions give the absolute observation IDs relative to the entire workfile range.

I'd like to know the observation IDs relative to the first observation within the sample.

[EViews Gareth]

I cannot think of an easy way to get it, no :(.

[startz]

Something like

Code: Select all

series foo = @obsid - @min(@obsid)

[EViews Gareth]

That assumes the observations in the sample are consecutive.

[jthodge]

Startz's idea should work for me since the observations within my sample are consecutive.

However, to clarify, I think you first need to set the sample in the workfile, of course, and you need to add 1 so that the first obs id within the sample is 1 and not 0:

Code: Select all

smpl ... series foo = @obsid - @min(@obsid) + 1

I didn't think about what would happen with non-consecutive observations in the sample. That would definitely add some complexity.

[EViews Gareth]

This works in the more general, non-consecutive case:

Code: Select all

series ser = 1 series id = @cumsum(ser)

Since ser will only be equal to 1 for observations in the current sample, the cumulation works.