[Talk-us] Non-Integer addresses

[David Lynch]

On Sun, Nov 15, 2009 at 13:02, Dave Hansen <dave at sr71.net> wrote:
> So, it seems that the TIGER data have some interesting addresses like:
>
> Non integer address: 9-35
> Non integer address: 9-01
> Non integer address: K200
> Non integer address: K210
>
> Anybody have any thoughts on how we should handle these?  The conversion
> script complains about them but I'm not even sure if it puts them in
> anyway or just warns.
>
> -- Dave

My first thought would be to go ahead and import them anyway and let
the users of the data figure out how to interpret them. Assuming the
first pair and last pair are each on the same segment, my intuition
would be that one refers to 9-01, 9-02, 9-03... and the second to
K200, K201, K202..., but I can't think of a good algorithm to match
these that wouldn't fail with the addresses Ian Dees mentioned that
have two independent integers in one alphanumeric string.

Just to add one more wrinkle, Austin has a few addresses that have a
fraction added to the end (e.g., down one side of a street, you might
have 402, 402 1/2, 404,) presumably from when one property was split
off from another.

-- 
David J. Lynch
djlynch at gmail.com