An innocent-enough question on the demographics of WMUK's membership has led me a merry dance so far. I'm sharing it with this list since it seems interesting in its own right, and (judging by chat at Sunday's meetup) someone may well know more than I do about it.
The idea is to start with postcodes, rounded off like CB2, for each member, and then do a graphical plot. Since the so-called postcode centroids are well known data, that part isn't hard. But the big urban areas will not be clearly represented.
The key word here seems to be "cartogram": see WP article. What I really would like is a "population cartogram", of area type, so a smoothing out of population density to be more even in a distorted map. There are examples as Figure 9 and Figure 10 in
http://www.statistics.gov.uk/downloads/theme_population/SMPS58.pdf
Those, though, are not based on postcode areas, but on census data done by wards.
Anyone (I'm sure we have some experts) on the list who understands the point here? Locating the postcode centroids within their census ward, once an for all, (i.e. just some matching that may have been done already) looks to be enough for the basic purpose of displaying membership data to find clusters. Alternatively some sort of population cartogram based on postcode areas could do it. Doesn't look like rocket science. Locating the postcode centroid by district so that the Figure 10 map of the PDF could be used as a basis could be done with the list on the next page (p. 24 of document).
Comments, please.
Charles
I have done a lot of geostatistics over the lasts three years so I could look into doing something although I wouldn't look at the distorted cartogram since I despise them with a passion :) Also doing it by Postcode centroids would be far too small for our purpose but looking at it at a council boundary level would be better and cuts the amount of work needed to get something useful :)
Seddon
On Fri, Dec 17, 2010 at 8:01 AM, Charles Matthews < charles.r.matthews@ntlworld.com> wrote:
An innocent-enough question on the demographics of WMUK's membership has led me a merry dance so far. I'm sharing it with this list since it seems interesting in its own right, and (judging by chat at Sunday's meetup) someone may well know more than I do about it.
The idea is to start with postcodes, rounded off like CB2, for each member, and then do a graphical plot. Since the so-called postcode centroids are well known data, that part isn't hard. But the big urban areas will not be clearly represented.
The key word here seems to be "cartogram": see WP article. What I really would like is a "population cartogram", of area type, so a smoothing out of population density to be more even in a distorted map. There are examples as Figure 9 and Figure 10 in
http://www.statistics.gov.uk/downloads/theme_population/SMPS58.pdf
Those, though, are not based on postcode areas, but on census data done by wards.
Anyone (I'm sure we have some experts) on the list who understands the point here? Locating the postcode centroids within their census ward, once an for all, (i.e. just some matching that may have been done already) looks to be enough for the basic purpose of displaying membership data to find clusters. Alternatively some sort of population cartogram based on postcode areas could do it. Doesn't look like rocket science. Locating the postcode centroid by district so that the Figure 10 map of the PDF could be used as a basis could be done with the list on the next page (p. 24 of document).
Comments, please.
Charles
Wikimedia UK mailing list wikimediauk-l@wikimedia.org http://mail.wikimedia.org/mailman/listinfo/wikimediauk-l WMUK: http://uk.wikimedia.org
On 17/12/2010 08:28, Joseph Seddon wrote:
I have done a lot of geostatistics over the lasts three years so I could look into doing something although I wouldn't look at the distorted cartogram since I despise them with a passion :) Also doing it by Postcode centroids would be far too small for our purpose but looking at it at a council boundary level would be better and cuts the amount of work needed to get something useful :)
Whatever. But the point is we have the postcodes, and the idea is to have a tool set up to display from that data alone. In fact I'm somewhat surprised that at least some visualisation tool of this kind isn't included in the open-source contact relationship manager we use.
Charles
Charles, I'm certainly interested. I would like to find out how may wikimedians there are in the midlands. I intend to launch an idea which is based here. At the momement I'm waiting for feedback from a number of leading wikimedians (hint yes you three), but I am unsure as to how many people are a small car ride away. So hope this idea works.
Seasons wishes Roger
On 17 December 2010 08:01, Charles Matthews <charles.r.matthews@ntlworld.com
wrote:
An innocent-enough question on the demographics of WMUK's membership has led me a merry dance so far. I'm sharing it with this list since it seems interesting in its own right, and (judging by chat at Sunday's meetup) someone may well know more than I do about it.
The idea is to start with postcodes, rounded off like CB2, for each member, and then do a graphical plot. Since the so-called postcode centroids are well known data, that part isn't hard. But the big urban areas will not be clearly represented.
The key word here seems to be "cartogram": see WP article. What I really would like is a "population cartogram", of area type, so a smoothing out of population density to be more even in a distorted map. There are examples as Figure 9 and Figure 10 in
http://www.statistics.gov.uk/downloads/theme_population/SMPS58.pdf
Those, though, are not based on postcode areas, but on census data done by wards.
Anyone (I'm sure we have some experts) on the list who understands the point here? Locating the postcode centroids within their census ward, once an for all, (i.e. just some matching that may have been done already) looks to be enough for the basic purpose of displaying membership data to find clusters. Alternatively some sort of population cartogram based on postcode areas could do it. Doesn't look like rocket science. Locating the postcode centroid by district so that the Figure 10 map of the PDF could be used as a basis could be done with the list on the next page (p. 24 of document).
Comments, please.
Charles
Wikimedia UK mailing list wikimediauk-l@wikimedia.org http://mail.wikimedia.org/mailman/listinfo/wikimediauk-l WMUK: http://uk.wikimedia.org
On 17/12/2010 09:45, Roger Bamkin wrote:
Charles, I'm certainly interested. I would like to find out how may wikimedians there are in the midlands. I intend to launch an idea which is based here. At the momement I'm waiting for feedback from a number of leading wikimedians (hint yes you three), but I am unsure as to how many people are a small car ride away. So hope this idea works.
It's a tricky counting exercise, though the answer may be "100 in round terms" anyway. More to the point is how to contact people ... going by categories, userboxes and so on doesn't tell you activity level.
Charles
I've worked with commercial suppliers of this sort of data, but don't know how much is available as freeware.
Council wards vary greatly in population, the boundary commission seeks to standardise them within individual authorities, but not nationally. Parliamentary constituencies are much more harmonised, though with anomalies especially where Islands are concerned.
There are about a 100 Postcode areas - the leading alphabetic bit. But they vary radically in population size - BT has three times the average and two of the Scottish islands are titchy - I think less than a tenth of average size.
Outward Postcodes (the first half of the Postcode) also vary widely in number of households.
Postal Sectors (the first half of the Postcode and the number at the beginning of the second half) are my favourite, the residential ones usually have around 2,000 households
WereSpielChequers
On 17 December 2010 10:08, Charles Matthews charles.r.matthews@ntlworld.com wrote:
On 17/12/2010 09:45, Roger Bamkin wrote:
Charles, I'm certainly interested. I would like to find out how may wikimedians there are in the midlands. I intend to launch an idea which is based here. At the momement I'm waiting for feedback from a number of leading wikimedians (hint yes you three), but I am unsure as to how many people are a small car ride away. So hope this idea works.
It's a tricky counting exercise, though the answer may be "100 in round terms" anyway. More to the point is how to contact people ... going by categories, userboxes and so on doesn't tell you activity level.
Charles
Wikimedia UK mailing list wikimediauk-l@wikimedia.org http://mail.wikimedia.org/mailman/listinfo/wikimediauk-l WMUK: http://uk.wikimedia.org
On 17/12/2010 18:47, WereSpielChequers wrote:
I've worked with commercial suppliers of this sort of data, but don't know how much is available as freeware.
Council wards vary greatly in population, the boundary commission seeks to standardise them within individual authorities, but not nationally. Parliamentary constituencies are much more harmonised, though with anomalies especially where Islands are concerned.
There are about a 100 Postcode areas - the leading alphabetic bit. But they vary radically in population size - BT has three times the average and two of the Scottish islands are titchy - I think less than a tenth of average size.
Outward Postcodes (the first half of the Postcode) also vary widely in number of households.
Postal Sectors (the first half of the Postcode and the number at the beginning of the second half) are my favourite, the residential ones usually have around 2,000 households
The interesting thing (if you could call it that) about this problem is that everyone I mention it to has a different take. It's the opposite of there being a solution known to everyone in the trade: suggestions are disparate. Which is not so good if all that is required is one solution fit for the basic purpose ...
Charles
On 17/12/2010 08:01, Charles Matthews wrote:
The key word here seems to be "cartogram": see WP article. What I really would like is a "population cartogram", of area type, so a smoothing out of population density to be more even in a distorted map. There are examples as Figure 9 and Figure 10 in
Super Output Areas?
http://en.wikipedia.org/wiki/Super_Output_Areas#Super_Output_Areas
Gordo
wikimediauk-l@lists.wikimedia.org