[VoiceOps] Rate Center Maps, Locations for NPANXXs

Peter Beckman beckman at angryox.com
Mon Sep 20 22:23:27 EDT 2010

On Mon, 20 Sep 2010, Troy Davis wrote:

> On Mon, Sep 20, 2010 at 11:08 AM, Peter Beckman <beckman at angryox.com> wrote:
>> In the past I've paid MelissaData.com $400 or so to get access to their
>> NPANXX-Ratecenter-Location-GIS data.  For each NPANXX, they list the
>> Ratecenter, City and state/region, Lat/Lon coordinates.  It's fairly
>> accurate, though I'm not entirely sure how they get the data.
> We ran the ratecenter through Google's geocoder, in some cases with a bit
> ofB preprocessing and/or filters (state/county where known).  But it's very
> much a best guess.  Our goal with Digits has been to expose everything we
> can provide for free. Sometimes that's quite reliable (rate center), some of
> which is mostly accurate (carrier, city), and some of which is a crapshoot
> (lat/lng).
> Those are cases of garbage in, garbage out.  We start with mediocre data
> (ratecenter string like "WSNGTNZN17") and run with it as far as possible.
> Some ("SEATTLE") are easier than others :-)
> I'd definitely consider replacing these lat/lng values with the CO lat/lng
> if you're able to find an unencumbered copy.  I think you're right that they
> would be more accurate.

  Actually I was saying in some cases it would be way inaccurate.  Take
  571-269.  The CO is listed as Beltsville, MD, but the rate center is
  listed as being in a different state entirely.  Using the CO lat/lng would
  confuse the issue.

  I used to think the CO lat/long would be more accurate, but I've learned

> As far as current accuracy, it depends on what you want to do with the
> data.  It's completely unsuitable for making call routing or rating
> decisions.  It's ideal for customizing/localizing IVRs, displaying inline
> maps, analyzing CDRs, deciding whether to offer a SMS menu choice, data
> appending, and the like, because all of those depend on depth of data and
> flexibility more than accuracy.

  Yep.  Though it would be great to try to get the City name normalized to a
  real city name, rather than NWYRCYZN01 for a ratecenter of NWYRCYZN15.

> Regarding John Todd's note that wireless users and VoIP carriers have
> decoupled number from location, a byproduct is that for non-routing uses,
> that makes Digits about as accurate as anything else -- not because Digits
> is dead on, but because nothing else is anymore either.  Digits might be 50
> miles off, but there's a good chance that a wireless subscriber is 50 miles
> from their area code center anyway, so the API consumer already expects
> imprecision (except for routing).
> Good catch, and thanks for the heads up.  I just added geo_precision to the
> docs and examples, and clarified its meaning and where lat/lng come from.

  Thanks for the clarification and link!

  I've started a wiki page on voip-info.org to try to put our collective
  heads together about regions and ratecenter names that don't map nicely to
  a city name.  Any other thoughts are appreciated.

  I'll make it prettier tomorrow.


Peter Beckman                                                  Internet Guy
beckman at angryox.com                                 http://www.angryox.com/

More information about the VoiceOps mailing list