Share
Embed
Some developments in the specification, estimation and testing of geodemographic classifications
13th April 2009, Kyoto, Ritsumeikan University
alexsingleton
April 13, 2009
Transcript
47 slides
-
SOME DEVELOPMENTS IN THE SPECIFICATION, ESTIMATION AND TESTING OF GEODEMOGRAPHIC
CLASSIFICATIONS Dr Alex D Singleton Department of Geography and Centre for Advance Spatial Analysis , University College London www.alex-singleton.com -
2006 DEFINITION OF A PLANET • After much debate the
International Astronomers Union (IAU) agreed… 1. The object must be in orbit around the Sun. 2. The object must be massive enough to be a sphere by its own gravitational force. 3. It must have cleared the neighbourhood around its orbit. -
FORMAL CLASSIFICATION 3 Main Divisions (Kingdoms) 1) Minerals 2) Vegetables
3) Animals Carl von Linné 17th Century Swedish Botanist -
BLACK: Lowest class. Vicious, semi-criminal. DARK BLUE: Very poor, casual.
Chronic want. LIGHT BLUE: Poor. 18s. to 21s. a week for a moderate family PURPLE: Mixed. Some comfortable others poor PINK: Fairly comfortable. Good ordinary earnings. RED: Middle class. Well-to-do. YELLOW: Upper-middle and Upper classes. Wealthy Charles Booth – Booth Map London 1889 -
WHY DO WE CLASSIFY? • Because the world is complex
• Need to simplify reality • Improve our understanding and help us make decisions • Help us to navigate the world • No representation is correct! -
WHAT ARE GEODEMOGRAPHICS? “Analysis of people by where they live”
or “locality marketing” (Sleight, 1993:3) Person Home Address “Area” -
HISTORY GEODEMOGRAPHICS? • Developed in 1970’s - attributed to Richard
Webber • Target urban deprivation funding : Identify clusters of similar neighbourhoods • Moved to CACI: Linked ED to Postcode • ACORN (Private Sector) • Moved to Experian • MOSAIC (Private sector) -
CREATING A GEODEMOGRAPHIC 46% 54% Census Non Census 70% 30%
Census Non Census Experian: Mosaic CACI: Acorn Census and Non Census -
CREATING A GEODEMOGRAPHIC • Age • Ethnicity • Country of
Birth • Population • Living Arrangements • Family Size • Tenure • House Type / Size • House Quality • House Ownership • Health of Population • Employment • Industry Sectors Census Only http://www.areaclassification.org.uk/ -
SEGMENTATIONS ARE CREATED BY CLUSTER ANALYSIS Area V1 V2 V3
V4 V5 V6 V7 V8 V9 V10 ... Area1 Area2 Area3 Area4 Area5 Area6 Area7 Area8 ... -
Variable 1 Variable 2 Cluster 1 Cluster 2 Cluster 3
CLUSTER ANALYSIS -
Area Cluster Area1 1 Area2 1 Area3 2 Area4 1
Area5 3 Area6 3 Area7 3 Area8 2 ... OUTPUT -
OUTPUT AREA CLASSIFICATION • 3 Levels – 52 Sub Groups
– 21 Groups – 7 Super Groups • Open Methodology – Only classification with status of a National Statistic • Peer Reviewed – Vickers, D.W. and Rees, P.H. (2007). Creating the National Statistics 2001 Output Area Classification. Journal of the Royal Statistical Society, Series A • Free! • RSS active user group -
1: Blue Collar Communitie s 1a: Terraced Blue Collar 1a1:
Terraced Blue Collar (1) 1a2: Terraced Blue Collar (2) 1a3: Terraced Blue Collar (3) 1b: Younger Blue Collar 1b1: Younger Blue Collar (1) 1b2: Younger Blue Collar (2) 1c: Older Blue Collar 1c1: Older Blue Collar (1) 1c2: Older Blue Collar (2) 1c3: Older Blue Collar (3) 2: City Living 2a: Transient Communities 2a1: Transient Communities (1) 2a2: Transient Communities (2) 2b: Settled in the City 2b1: Settled in the City (1) 2b2: Settled in the City (2) 3: Countryside 3a: Village Life 3a1: Village Life (1) 3a2: Village Life (2) 3b: Agricultural 3b1: Agricultural (1) 3b2: Agricultural (2) 3c: Accessible Countryside 3c1: Accessible Countryside 3c2: Accessible Countryside -
These neighbourhoods are predominantly located in inner cities and are
characterised by low quality, high density rented flats. They have multi-ethnic populations, many of whom are first generation immigrants. Other residents include students. -
MOSAIC • 2 Levels • 61 Types • 11 Groups
• Closed Methodology • Unit Postcode • Commercial • Just data or analysis tools -
Urban Intelligence Young, single and mostly well educated, these people
are cosmopolitan in tastes and liberal in attitudes. Group 5 / 11 -
Symbols of Success Happy Families Suburban Comfort Ties of Community
Urban Intelligence Welfare Borderline Municipal Dependency Blue Collar Enterprise Twilight Subsistence Grey Perspectives Rural Isolation Cheltenham, Gloucestershire, UK. -
WHAT DO WE TAKE FROM THIS? • Not all segmentations
are the same… • OAC says society is made up of 52 different types of areas? • MOSAIC says society is made up of 61 different types of areas? • Which is right? • Commercial Providers will more than likely say: Ours is! -
RECENT “INNOVATIONS” • Commercial innovation - diversification / new markets
• Region: Global, London, Scotland, Northern Ireland, Japan, Italy... • New Industries: Automotive, Financial, Grocery, Public Sector, Health • New Segments: Daytime, Names, Income, Street Values, Shareholder Activity, Unemployment, Individuals (Scale) -
“RECENT” INNOVATIONS • Long history of academic critique • Fuzzy
Geodemographics (Feng & Flowerdew, 1998; See & Openshaw, 2001) • Description V Modelling (Harris et al, 2007) • Bespoke Classifications (Longley and Singleton, 2009a) • Public Consultation (Longley and Singleton, 2009b) • Utility of public consultation data (Singleton, 2008) -
“RECENT” INNOVATIONS • Need for real time geodemographics: • Current
classifications are created using static data sources. • Rate and scale of current population change is making large surveys (census) increasingly redundant. • Significant hidden value in transactional data • Data is increasingly available in near real time • Application specific (bespoke) classifications have demonstrated utility. -
REALTIME FEEDS OF DATA • Involve integration of large and
possibly disparate databases • E.g. Doctor registrations; census data • Common protocol • XML: E.g. UK Neighbourhood Statistics API • Is there any value in other non- traditional data sources? • E.g. Flickr -
ONLINE SPECIFICATION INPUTS • Usability • Expert V Non-Expert Users
• Non-Expert Users • Pre selection of variables and weighting for specific application • Expert Users • Selection of any variable and any weighting -
CLUSTERING • K-Means algorithm: Unstable - initial start conditions effect
the results • Measured within sum of squares or R2 K=4 -
CLUSTERING • Alternate algorithms? • PAM (Partitioning around medoids) tries
to minimize the sum of distances of the objects to their cluster centers. • CLARA draws multiple samples of the dataset, applies PAM to each sample and returns the best result. • GA (Genetic Algorithm) is inspired by models of biological evolution. Produce results through a breeding procedure. -
CLUSTERING K-means result for 41 “OAC variables” K-means result for
26 OAC Principle Components 99% Similar K=4 or... refine k-means -
STATE OF THE ART • Slowly moving beyond: • Idea
of expert producers • General purpose classifications • There is only one correct representation • Creating classifications which are • Responsive to changes in local populations • Fit for purpose (bespoke classifications) • Open to scrutiny and verifiable by the public