Finding common last observation in workfile

[javiersan]

Hello,

I wonder if there is a way of finding the last date that has a non-na observation for all the series in a workfile (or group).

At the moment this can be achieved by the following convoluted commands:

Code: Select all

group g * series tracklast=@recode(@rnas(g)=0,na,@rnas(g)) !lastv=@ifirst(tracklast)-1 %last_dt=@otod(!lastv) smpl {%last_dt} {%last_dt}

Is there a more elegant code that achieves the same result?

Thanks,

Javier

[javiersan]

Actually, the code only works if the series have no gaps, so for most cases it may have to be modified.

Javier

[trubador]

There is more than one way to do that. Here is one:

Code: Select all

!nseries = my_group.@count vector(!nseries) datevals svector(!nseries) datestrs for !i = 1 to !nseries %name = gr.@seriesname(!i) datestrs(!i) = {%name}.@last 'Dates will be stored in string format and you can keep this as a list for further use datevals(!i) = @dateval({%name}.@last) 'Dates will be stored in number format to find the latest one next %last = @datestr(@max(dates)) smpl {%last} {%last}

[EViews Gareth]

I'm posting from my phone, so can't test this, but something along the lines of:

Code: Select all

smpl if @robs(g)>0 series t = @trend+1 string date = @otod(@max(t))

[javiersan]

It does work with the following variation:

Code: Select all

smpl if @robs(g)=g.@count series t = @trend+1 string date = @otod(@max(t))

Thanks,

Javier

[EViews Gareth]

Ah, yes, I misunderstood whether you wanted the last date for which any series has a value or the last date for which all series have a value.

[EViews Glenn]

Or more directly

Code: Select all

smpl if @rnas(g)=0