Re: st: Interactions between tsset and sort

[Nick Cox <njcoxstata@gmail.com>]

I'd suggest that the minimum for a -*set- command would allow
declaration of identifier variables and optiionally a preferred sort
order. Off the top of my head some common types of spatial data are

area           name_identifier (location information held indirectly if at all)
point_2d     x_coordinate, y_coordinate
point_3d     x_coordinate, y_coordinate, z_coordinate
line            from_identifiers, to_identifiers (arbitrary)
flow           from_identifiers, to_identifiers (NOT arbitrary)

and I'd recommend that you choose one of two paths:

1. You code for yourself in the spirit of "If you have the same
problem, you are welcome to use my code".

2. You code very openly allowing users to set any number of identifier
variables and to declare a type of spatial data that you have not even
thought of or encountered yourself.

Nick

On Sat, Feb 12, 2011 at 3:58 PM, Gordon Hughes <G.A.Hughes@ed.ac.uk> wrote:
> Thank you for both suggestions.  I will consider both of them.  Two remarks:
>
> A.  The manual entry for xtset states that
>
> "One of the syntaxes of tsset is tsset panelvar timevar, which is identical
> to one of xtset's syntaxes, namely, xtset panelvar timevar.  Here they are
> in fact the same command ..."
>
> I think that the --xtset-- behaviour of not sorting the data may only apply
> when you use --xtset panelvar-- which is not sufficient for my purpose as I
> need to identify the time variable as well.  I had taken the manual
> literally, but I may have misinterpreted it.
>
> B.  Writing a modified version of tsset (xtset) - e.g. spatset - would offer
> the possibility of setting data up for other kinds of spatial analysis, so
> it would be interesting to find out what features potential users might
> regards as being useful.

*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/