Postal Code Distance

P

pilks

Well-Known Member
Joined
Oct 9, 2009
Messages
129
Reaction score
0
We hired some interns over the festive season to do some data capturing for us. They had to capture some addresses into a form and then centre the map on the location of the address. Needless to say, Interns being interns got this wrong and now we have some addresses that are in the wrong suburb and others in different cities. I have been tasked in finding all the addresses who Geo Co-ordinates don't match up/come close with the address.
The table I have is
Address:
Street line 1
Street line 2
Suburb
Postal Code
Town
Province
Longitude
Latitude

I don't want to do a manual audit through 1000s of records.
I had thought of doing the following:
1. Look up all the postal codes in all the records I have
2. Get the Co-ordinates for those
3. Calculate the distance between the actual postal code co-ordinate and the co ordinate we have stored
4. If distance exceeds X flag.

I have a few pondering points
1. How big must I make X. I was thinking 20 km. But it could be less.
2. Is there a smarter way to do what I am planning to do ?

Thanks
 
You could also write a small application that feeds Google Maps the addresses and returns the longitude and latitude for you? Depends on the location and quality of the addresses though as I find Google Maps struggles with finding addresses in certain areas.
 
Messugga said:
You could also write a small application that feeds Google Maps the addresses and returns the longitude and latitude for you? Depends on the location and quality of the addresses though as I find Google Maps struggles with finding addresses in certain areas.
Click to expand...



Ye I have found that too with the previous lot of addresses we feed in to the system, so i would prefer not to rely on Gmaps for exact Geo co - ordinates. I need to use a distance
 
/writes script to scrape results from sources, no interns needed. Just coordinates need manual entry for exact point.
 
Maybe get another set of interns to capture the data again and then only look at the ones that are significantly different.

Or the Google maps option above could maybe eliminate/verify the ones that are relatively good, so then you only have to check the remainder.
 
I would also go the google maps route. Then flag how many addresses that you cannot match. If the number is small enough, manually look at them.

If its too large, look into getting the data recaptured or programatically loaded in.
 
Get the list of post office postal codes (you sms your email address to a shortcode and then they email it to you immediately, costs something like R7, check their site).

Then filter all the duplicate ones (the original list is 15K but there are only a couple of thousand unique ones). Then pass this information to the google map API and cache the results.

Fortunately the list is small enough that you want hit your daily API call limit.

I did this to test a proposal we had for a client who had a site that allowed people to advertise rooms to rent/want and we had to match them up.
 
You must log in or register to reply here.
Back
Top
Sign up to the MyBroadband newsletter
X