Geodemographics 2.0

1
Geodemographics 2.0

• Research Challenges for Real Time Decision
  Makingin Public Sector Service Delivery

Alex Singleton pontificates about geodemographics
for about an hour...
2
Introduction

• Why Classification?
• An Evolving Research Themes
• Commercial
• Extension Strategies Rebranding,
  Diversification
• Academic
• Consumption
• Critique Production
• Geodemographics 2.0
• Key Challenges
• Geodemographics 2.0
• Relevance in a Depreciating Data Economy
• Complexity Time, Space and Scale
• Geocomputation Challenges

3
Why Classification?

• The world is complex
• Simplify Reality
• Improve our Understanding
• Help make decisions
• Help us to navigate the world

4
Evolving Research Themes
Critique / KT ?
Production
Consumption
Expertise
5
Evolving Research Themes

• Missing Knowledge Transfer?
• 2 Sides to a coin - Different Drivers

6
Commercial Innovations?

• What are the....

7
An Evolving Research Agenda

• Commercial Innovations
• New Markets Diversification

One Size Fits All?
New Segments Daytime Names Income Street
Values Shareholder Activity Unemployment Individua
ls (Scale)
Region Global London Scotland Northern Ireland
New Industries Automotive Financial Grocery Public
Sector Health
8
An Evolving Research Agenda

• Commercial Innovations Extension Strategies
• Rebranding

Market Share
Time
9
(No Transcript)
10
Academic Innovations?

• What are the....

11
An Evolving Research Agenda

• Academic 10 Years Critique Production

London OAC Health OAC Education OAC E-Society
2001 Census
1991 Census
GBProfiles (Openshaw)
OAC (Vickers)
LSOAC (ONS)
Production
Live Classification Updates Singleton and
Longley (2009b)
Public Consultation Longley and Singleton (2008)
Bespoke Geodemographics Longley et al (2008)
Singleton and Longley (2009a)
Methodology Application
Description V Modelling Harris et al (2007)
Fuzzy Geodemographics Feng and Flowerdew (1998)
Fuzzy Geodemographics See and Openshaw (2001)
Diversity of Diversity Voas and Williamson (2001)
Ground Truth Goss (1995a, b)
Consumption Reader Goss (2003)
Critical Theory
Time
12
Critical Theory

• Arguments Response / Lessons
• Monitor, Model and Control
• Consumers are marketed identities
• Assumes consumers are not rational
• Strategy Metaphor Simplify and Mystify
  technology purpose gaining market share
• Relates to products essentially using the same
  data just a different badge.
• True, there is little to choose between the
  products
• Argument demonstrates further contempt for
  consumers!
• Cluster Analysis Aggregation Errors
• True, however doesnt entirely mitigate the value
• Assumes Social identity perfectly correlates with
  consumption
• Good point bespoke classifications!

13
Methodology and Application

• Fuzzy - Discrete V Continuous Representation
• Good idea communication a problem
• Sensitivity to local conditions
• True, compress multiple space time dimensions
  into core components
• Modelling versus Description
• Yes!
• Application Specific (Bespoke)
• Yes, but how do these become flexible and
  embedded for general use.
• Consumer Feedback Challenge Assignmets
• Sounds good!
• Building Feedback into Classification
• How are these verified?

14
Key Challenges?

• What are the....

15
Everything 2.0
Academic
Critique / KT ?
Consumption
Commercial
16
Challenges for Everything 2.0

• Changing relationships with space time (new
  channels)
• Rising Importance of Virtual Places (Social
  media, social networks)
• Do individuals interact in real places in the
  same way as virtual places? How do we reach them?
• Changing responses to information
• Consumption (Encarta) V Production (e.g. Wiki)
• Static V Fluid
• Clever Consumers
• Information savvy - B2B Public
• Data Economy Paradox
• Data Economy has never been stronger!
• Data Economy has never been more restricted?
• Lead datasets (electoral roll) increasingly
  marginalised

17
Geodemographics 2.0

• What is my response....

18
Academic Role in Geodemographics 2.0

• Commercial providers will operate business as
  usual until something better comes along. Why
  rock the boat?
• E.g. Legacy of K-Means - RIP
• Little engagement with any critique
• Some evidence _at_ CACI (e.g. Health Acorn)
• IMPORTANT
• PARTNERSHIP KNOWLEDGE TRANSFER!

19
Geodemographics 2.0
Bespoke Requests
Realtime Measurement
Reality
20
Geocomputation Challenges

• Data
• Variable Relationships Verification
• Spatio-Temporal Sensitivity
• Processing Analysis
• Dissemination

21
Data

• Individual Level Data
• Confidentiality Disclosure Control
• Multiple Sources Data Protection
• Area Level Data
• Managing different scales safely
• Disparate Data
• Standards XML
• Secure - Encryption
• Speed of Transfers

22
Relationships Verification

• Assuring Integrity Validity
• Ontological and Meta Information
• E.g. How old is the data?
• Which data is behavioural?
• Which data is attitudinal?
• What is the coverage?
• Time, Place
• Interpreting user requests
• Which variables?
• Which weight?
• Spatial Weights
• Temporal Weights

23
Spatio-Temporal Sensitivity

• Accounting for local variation
• Spatial Interaction Models?
• Geographically Weighted Cluster Analysis
• Temporally Weighted Cluster Analysis
• Normalisation which is sensitive, without
  detrimentally effected by outliers

24
Processing

• Which clustering algorithms?
• Not K-Means
• Not stable, cant efficiently optimise why do
  we use them?
• Maybe Partitioning Around Medoids (PAM)
• Maybe Intelligent Genetic Algorithms
• Maybe something entirely different
• Building for a single purpose
• Multidimensional Scaling
• Principle Components
• Processing which doesnt collapse the geographic
  dimension of the data
• Processing which is FAST!
• Safe to use Testing!

25
Dissemination

• Live Classifications require...
• Live Pen portraits
• Build automated descriptions for clusters based
  on appended data
• Live Maps
• Automated Google Map Generation
• Overcome tile processing limitations
• Live Images
• Automated production of appropriate pictures and
  collage imagery

26
Geodemographics 2.0
Geodemographics 2.0
Academic
KT
Commercial
27
References

• Feng Z and Flowerdew R, 1998, Fuzzy
  geodemographics a contribution from fuzzy
  clustering method. In Carver S (Ed.), Innovations
  in GIS 5. London Taylor and Francis, pp.119127
• Goss, J. 1995a Marketing the new marketing the
  strategic discourse of geodemographic information
  systems. In Pickles, J., editor, Ground truth,
  New York Guilford Press , 130-170.
• Goss, J. 1995b We know where you are and we
  know where you live the instrumental
  rationality of geodemo-graphic information
  systems . Economic Geography 71, 171-198
• Goss, J. (2003) We know who you are and where
  you live The Instrumental Rationality of
  Geodemographic Information Systems. In David B.
  Clarke, Marcus A. Doel, Kate M. L. Housiaux The
  Consumption Reader. Routledge.
• Harris R, Johnston R, Burgess S, 2007,
  Neighborhoods, ethnicity and school choice
  developing a statistical framework for
  geodemographic analysis, Population Research and
  Policy Review, 26, 553-579.
• Longley, P.A., Webber, R., Li, C. (2008) The UK
  geography of the E-Society a national
  classification. Environment and Planning A. In
  Press.
• See, L. and Openshaw, S. 2001. Fuzzy
  geodemographic targeting. In Regional Science in
  Business, Clarke, G. and Madden, M. (eds).
  Springer, Berlin. pp.269-282.
• Singleton, A., Longley, P.A. (2009a) Creating
  Open Source Geodemographics - Refining a National
  Classification of Census Output Areas for
  Applications in Higher Education. In Press Papers
  in Regional Science.
• Singleton, A., Longley, P.A. (2009b) As yet
  untitled Area paper Relates to updating
  classification.
• Longley, P., Singleton, A.D. (2008)
  Classification through Consultation public views
  of the geography of the e-Society. International
  Journal of Geographical Information Science. In
  Press.
• Voas, D. and Williamson, P. (2001), The diversity
  of diversity a critique of geodemographic
  classification, Area, 33 (1), 63-76.