Great job what a fantastic resource. One quick question i have is can i use this data in order to work out radius positions form a give postcode. ie if i had 200 property postcodes around the uk and then i get a user to enter theres i need to say get all properties within 30/40/50 miles etc of that postcode. If it can can anybody advise how or even if this data does not go down to the next level needed is there any where i can get this data. Obviosuly PAF is not an option.

Thanks all in advance

This seems to be a great resource so congratulations on doing such a great job.

I was hoping someone could shed some light on a requirement i have. Ok i have a list of lets say 300 houses all over the uk which i have postcodes for. What i need to do is allow a user to enter there postcode and then display the houses within a certain radius ie 10,20,30 miles etc. This resource i dont think does not narrow things down enough, if it does please say so, if not please does anybody know where i can get hold of it, PAF is not an option obviously. I will be doing this in asp.net if anyone can help me.

Thanks Jonathan

Do you or anyone else know of any resources that give the eastings/northings for Postal Towns? (So for example, when I search on Google Maps for Birmingham, Google takes me to the centre of Birmingham. I assume they have some way of working out the coordinates of the centre of a town?)

Is there a free resource that lists the central eastings/northings coordinates of a town?)

——————————————–
Dear Mr Harrison,

Thank you once again for contacting us with regards to our free centroid databases. We have carefully considered all of your points. In order to calculate a true centroid we have to equate for every geographical coordinate, regardless of whether it is associated with a po box number or a residential or commercial property. Non-unique locations are not always associated with po box numbers. We have taken a look at the B1 district area and were very surprised to find that the centroid given had so much weight pulling it towards Bath. It does not match our own calculations. After analysing the code we use to generate these databases online, we realised that the reason for the discrepancy was because of far too much duplicate weighting factors. In other words we had been feeding in duplicate coordinates into our equations and that resulted in what we refer to as biased or unfair weight. We must apologies for this. The problem was due to issues porting our original delphi code over to php for use online. This has now been addressed. All calculations are now based on unique coordinates, which is as they should be. A single coordinate that is located a great distance away from the majority of district or sector locations will only ever have a very negligible influence concerning the location of the true centroid coordinate. This is another reason why ignoring non-unique locations altogether would be a very bad idea, as they do not actually have much sway over the position of the final centroid, when you consider the weighting factor in its entirety. Also, this may or may not be of interest to you, but we now offer the same databases with average centroid calculations. This works by taking the average coordinate as the centroid coordinate. Both databases can be generated and downloaded from our website here: http://www.postcodepal.com/?page=database-generator.

If you have any further queries then please do not hesitate to contact us.

Yours Sincerely,
Mark Johnston
——————————————–

Not quite sure what advantage an average centroid has over a true centroid?

Dean

If you would like to sell it please contact me on: firma(at)aycan.co.uk

Thanxs

Blindly calculating district/sector centroids from all raw postcode locations doesn’t work very well in many cases.

Example;

“B1″,”Birmingham”,”403511″,”276448″,”52.385675″,”-1.948398″

If you plot this location on Google maps you can see that it is miles away from Birmingham. The reason for this is easily seen if you look at the raw Code-Point data. There are many postcodes within B1 with the location 375350,164483 (51.378366,-2.354215) – which is in the middle of Bath. (Although strangely, if you type one of these postcodes into Google maps, e.g. B1 3PF, it does show it in Birmingham…not sure where Google are getting this from).

Now, I don’t have access to the full Code-Point dataset so this is speculation, but seeing as all these postcodes share EXACTLY the same location I would guess that they are PO Box codes (the Code-Point Open data doesn’t include the PO Box indicator flag). In my opinion these codes should not be mixed with “real” codes in centroid calculations, because they are useless for just about any practical location usage scenario. If you want to know whereabouts of the B1 district then generally you don’t care about PO Box addresses in Bath.

So lacking PO Box indicators, what I do when processing the Code-Point Open data is to just ignore any non-unique locations. This works well in most cases, my result for B1 centroid is 406194,286859 (52.479252,-1.908789), which is within 100m of what Dan has.

Hope that helps

Cheers!

I’ve found that the locations in the PostcodePal files above are very suspect. They don’t match up with the Code-Point Open data very well, and some just look plain wrong.

As a quick example look at the very last district, ZE3. Code-Point puts all ZE3 postcodes (there are only 12 of them) on the southern tip of the island, but PostcodePal puts the ZE3 centroid in the middle of the island. I have no idea how they calculated that.

The town, district, county, and region data does seem to be accurate though.

This was discovered on wiki leaks and can be downloaded from the famous bit torrent site, infact the link is on wikilinks, it contains full postcodes etc, except for the damn street names, counties and towns.

It would be great if someone could put this together with the missing info, bewarned its a massive file, open in access etc as excel wont load all the entries!

http://postcodepal.com/?page=database-generator

The database links are

http://postcodepal.com/dbgen/postcode_districts.zip
http://postcodepal.com/dbgen/postcode_sectors.zip

I’m not interested in paying for a full database (can’t afford it) nor do I have the space to host it. The above is more than what I needed. Might save you some time and these ones are free I haven’t tested them for accuracy yet. Hopefully get some time this weekend.

All the best,
Dean

http://api.postcodepal.com/E138HR.xml
http://api.postcodepal.com/E138HR.json

Unfortunately it only accepts full post codes. Maybe it will help someone?

Regards.
Dean

All the best.
Dean

I’ll keep posting updates here as an when you release new versions.

Cheers Rob.

*** A week left before Open Data consultation closes ***

Can you go to http://data.gov.uk/opendataconsultation/questions#comment-7139
and make your voice heard too.

Thanks,
Tom

and then the main table, uk_postcodes which references them.

country_id, country_code_id, region_id and town_id in this table refer to each of the above reference tables respectively.

The file is in pg_dump format (easily ported to other DBMSes). Replace YourSchemaOwner to the owner of the schema you are loading into (if public then this is usually the database owner)

Download from here: http://www.sms-speedway.com/uk_postcodes_05.pgdump.bz2

http://www.dangibbs.co.uk/files/postcodes/dg-postcode_db-0.5.tar.bz2
http://www.dangibbs.co.uk/files/postcodes/dg-postcode_db-0.5.zip