Login  Register

Re: Zip codes and Electoral districts

Posted by Stéphane Zagar on Feb 09, 2007; 2:18am
URL: http://civicaccess.416.s1.nabble.com/Zip-codes-and-Electoral-districts-tp865p870.html

Daniel :

I don't see how this can be copyrighted. Someone (you Daniel ?) told that facts can't be copyrighted. I just can't imagine how this could be copyrighted ! Damned I could call every and each Canadien citizen and ask him his zipcode and his electoral district, then can't use this ?

Concerning the other points you raise :
- It's amazing that postal codes can overlap several districts... but I can't see how to solve this. Even the Election Canada website uses the postal code to find your district.
- I don't see how you get only 1 millions possibilites with this regex. You bring the first character from 26 letters to 19. The result is 19*10*26*10*26*10 = 12 millions


Franck :

I understand you point. However I still don't understand why those data are not freely accessible. If they are not available I'll pick them... if I can ;)

Steph

Daniel Haran wrote:
Hi Stéphane,

There are a few problems with that approach. First of all, the
lat/long mapping databases tend to be copyrighted themselves. Besides,
those are the centroids - the center-most point of a polygon.

One of the main problems Russell mentioned (and explained further at
http://www.digital-copyright.ca/node/1607 ) is that the mapping is not
one to one. A postal code region can overlap several electoral
districts.

The drupal lobby module he mentioned has a good regular expression for
validating postal codes:
/([ABCEGHJKLMNPRSTVXY]\d[A-Z]\s?\d[A-Z]\d)/i

For non-techies, that first segment ABC...Y are all the allowable
first letters in a postal code. That reduces the number of
possibilities to 1,216,800 from 17,576,000. At a second a piece,
that's a bit over 14 days from 100 :)

cheers,

Daniel.

On 2/8/07, Stéphane Zagar [hidden email] wrote:
  
Hey

In fact, technically it's not complicated to get a zip code mapping with
latitude and longitude :

H3T1A1;45.5113;-73.6162
H3T1B1;45.5083;-73.6183
H3T1C1;45.5016;-73.6245
H3T1E1;45.4974;-73.628
H3T1G1;45.51;-73.614
H3T1H1;45.5061;-73.6176
H3T1J1;45.4945;-73.6282
H3T1K1;45.5002;-73.6212
H3T1L1;45.4962;-73.6253
H3T1M1;45.4971;-73.6225
H3T1N1;45.501;-73.6178
H3T1P1;45.4942;-73.6229
H3T1R1;45.5105;- 73.6154
H3T1S1;45.5078;-73.6178
H3T1T1;45.5052;-73.6198
H3T1V1;45.5005;-73.6192
H3T1W1;45.4987;-73.6208
H3T1X1;45.4971;-73.6212
H3T1Y1;45.4959;-73.6206
H3T1Z1;45.4977;-73.6257

And I think it's not that much complicated to do the same reverse
engineering for electoral district.

The problem is that mathematically, there are about 17 millions of valid
combination of zip code. It take about 1 sec for my script to retrieve the
zipcode/lat-long so if we do a basic computation it gives us about... 100
days to retrieve everything (and probably a lot of bandwidth). Even if we
set the last number to 1 (what I did in my previous test), it's still about
2 millions valid values. And I'm not sure if the website I use to retrieve
this will allow this (with 1 request/sec, I can't imagine they won't see
that something is happening).

By the way, I have a question for geostuff people (Tracey ?) : On the
geogratis website, there are some files concerning the electoral districts (
http://geogratis.cgdi.gc.ca/geogratis/en/option/select.do?id=1169),
but I don't know anything concerning the format (Arc Export or shape file).
Is there a way to do something with that ? For example do a process
what-ever to find out in which electoral district is a point (long-lat) ?

Steph

_______________________________________________
CivicAccess-discuss mailing list
[hidden email]
http://civicaccess.ca/mailman/listinfo/civicaccess-discuss_civicaccess.ca


    

_______________________________________________
CivicAccess-discuss mailing list
[hidden email]
http://civicaccess.ca/mailman/listinfo/civicaccess-discuss_civicaccess.ca