UCL DEPARTMENT OF GEOGRAPHY Open Geodemographics: Open Tools and the 2011 OAC UCL Department of Geography, Gower Street, London, WC1E 6BT Chris Gale * Muhammad Adnan Paul Longley mapblog.in gis-tech.co.uk paul- longley.com @geogale @gisandtech * Conference attendance kindly supported by RGS-IBG funded QMRG bursary
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
UCL DEPARTMENT OF GEOGRAPHY
Open Geodemographics: Open Tools and the 2011 OAC
UCL Department of Geography, Gower Street, London, WC1E 6BT
Chris Gale* Muhammad Adnan Paul Longleymapblog.in gis-tech.co.uk paul-longley.com@geogale @gisandtech
* Conference attendance kindly supported by RGS-IBG funded QMRG bursary
UCL DEPARTMENT OF GEOGRAPHY
• What is Geodemographics?• Need for Open Geodemographics• GeodemCreator• The 2011 Output Area Classification• Summary
Outline
UCL DEPARTMENT OF GEOGRAPHY
Geodemographics• The analysis of people by where they live• Areas can be described by the characteristics and
attitudes of those people who live in them• Based on the concept that similar people with similar
characteristics are more likely to live within the same locality and that such area types will be distributed in different locations across a geographical space
• Commercial (MOSAIC, ACORN) and free (OAC) classifications available
UCL DEPARTMENT OF GEOGRAPHY
Commercial Geodemographic Classifications• Created as ‘black box’ systems (Longley and Singleton, 2009)• Closed methods are used with little documentation• Little information is given regarding the data inputs,
normalisation and weighting procedures, and clustering methods employed
UCL DEPARTMENT OF GEOGRAPHY
Need of Open, Transparent, and Flexible Classifications• Increased amount of data sources due to ‘open data’
initiatives– ONS NeSS data exchange, London data store, Crime data API
• Need of open methods– Open method of Estimation, Normalisation, and Clustering
procedures• Open public consultation
UCL DEPARTMENT OF GEOGRAPHY
Need of Open, Transparent, and Flexible Classifications• A number of statistical packages could be used for
building geodemographic classifications– R, SPSS, Microsoft Excel
• No unified software utility exists that could be used for building open, transparent, and flexible classifications
• ‘GeodemCreator’ is a unified software utility for building geodemographic classifications
UCL DEPARTMENT OF GEOGRAPHY
GeodemCreator• A cross platform java software utility for building
geodemographic classifications• Requires ‘Java’ and ‘R’ installed on user’s machine• Geodemographic classifications could be created for
any geographical level and by using any data set• Users can combine census data with their own data
sources
UCL DEPARTMENT OF GEOGRAPHY
GeodemCreator• Operates in ‘Basic’ and ‘Advanced’ modes
– Basic Mode is for inexperienced and new users– Advanced modes is for experienced users
• Clusters the data by using k-means clustering algorithm
UCL DEPARTMENT OF GEOGRAPHY
UCL DEPARTMENT OF GEOGRAPHY
GeodemCreator Case Study• A Socio-economic and Ethnic classification of Greater
London• Created by using 41 OAC variables and 12 ethnicity
variables (created from ethnicity data source http://worldnames.publicprofiler.org)
• GeodemCreator was used for building the final classification
GeodemCreator Case Study Results• A Socio-economic and Ethnic classification of Greater
London:
UCL DEPARTMENT OF GEOGRAPHY
GeodemCreator Case Study Results• GeodemCreator also produces radial charts for each
cluster solution
English and European ethnic groups living in suburban areas
Well off and educated Asian families
UCL DEPARTMENT OF GEOGRAPHY
GeodemCreator Case Study Results
English, European, and Celtic fringe city commuters
Poor Asian Families
UCL DEPARTMENT OF GEOGRAPHY
GeodemCreator Case Study Results
Childless European city dwellers Native blue collar communities
UCL DEPARTMENT OF GEOGRAPHY
GeodemCreator Case Study Results
English and European ethnic groups living in council properties
UCL DEPARTMENT OF GEOGRAPHY
The 2001 Output Area Classification (OAC)• Groups the UK population
into:– 7 Supergroups– 21 Groups– 52 Subgroups
• Only data source used is the 2001 Census– 41 Variables
• Variety of organisations use it including local government and commercial companies
UCL DEPARTMENT OF GEOGRAPHY
• Building on the success of the 2001 OAC• The 2001 OAC’s real achievement was showing that
open-source geodemographic classifications were possible
• Can utilise developments in computing over the past 6 years, since the 2001 OAC’s publication, to make improvements
• Can be produced using open-source software (if required) with a fully open and transparent methodology
The 2011 Output Area Classification
UCL DEPARTMENT OF GEOGRAPHY
• Not just a repeat of the 2001 Output Area Classification
• Methodology that will possibly not rely on 100% Census data
• Enhanced outputs to cater for different potential users• Designed to allow easy creation of bespoke variants
– Variables and/or Geography – Automated variable selection depending on user criteria
• e.g. variables used for a national classification not necessarily being suitable for a regional classification
The 2011 Output Area Classification
UCL DEPARTMENT OF GEOGRAPHY
• Code used to auto-select best variables for desired purpose
• Allows for a fully transparent and repeatable methodology– Variable selection the only “black box” element of the 2001
OAC• Allows for wider scale bespoke geodemographics
– A user with no geodemographics experience can produce their own classification by selecting the variables, standardisation method, number of clusters.
– Removes any technical barriers that could prevent wider adoption of bespoke geodemographic classifications.
2011 OAC Variables
UCL DEPARTMENT OF GEOGRAPHY
Bespoke Geodemographic Classifications• Categorised into 3 main types:
– Using the same data already provided in classification.– Changing the number of variables used to create a
classification.– Uploading other data that was not originally included into a
pre-existing classification or creating a new classification from scratch.
• In the case of OAC this could resolve a problem when used at a regional level– London is an example of one such region that OAC does not
classify very well.
UCL DEPARTMENT OF GEOGRAPHY
UCL DEPARTMENT OF GEOGRAPHY
• Bespoke free area classification of Hull
• 45 Census Variables used
• 10 Groups in 3 hierarchies
The Hull City Council Classification
UCL DEPARTMENT OF GEOGRAPHY
2011 OAC and Open Data• Would it be better to use potentially “newer” Open
Data (when compared with the 2011 Census)?• How much of a problem is the lack of data currently
available at OA level?• Using Open Data raises a lot of questions:
– What sources of Open Data should be used?– What should the coverage of the Open Data be?– Does the integrity of the Open Data matter?– How often should the Open Data sources be updated?
• Beyond 2011
UCL DEPARTMENT OF GEOGRAPHY
On-The-Fly Clustering• To meet the changing and varying needs of users a
dynamic classification environment needs to be created
• Ability to create bespoke classifications a requirement– both for different geographies (e.g. London or UK) and the
range and number of variables utilised (e.g. Census and/or non-Census) with an additional weighting capacity
• Will require clustering to happen in real-time• Research of users specific has been undertaken
– 2011 OAC User Engagement (run in partnership with the ONS)