Adding Value to the Census, March 2004
Seminar report
One of the principal ways of adding value to the Census lies in using
the results to build geodemographic classifications. Although the main
outputs from the 2001 Census were published barely six months ago, the
CGG seminar on 3rd March 2004 brought together experts from no less
than eight commercial developers to discuss their new systems. Around
seventy interested users attended in order to learn about the new products
and question the speakers.
The event was chaired by Keith Dugmore, from Demographic Decisions,
with considerable expertise. As the speakers were keen to provide as
much information as possible in their presentations, Keith also did
a sterling job of ensuring that the programme ran precisely on time.
The seminar began with two scene-setting talks. The first, by Peter
Sleight of Target Market Consultancy, presented the audience with a
roadmap of all the types of products available from 2001 Census data,
and so helped to position the presentations that followed. The second
briefing was given by Chris Denham and John Charlton from the Office
for National Statistics. Chris summarised very clearly the main features
of the data and how users can access it free of charge. Then John described
the area classifications that his team at ONS have built from the Census,
and demonstrated the interactive analysis and mapping available on the
Neighbourhood Statistics website.
After a short break, Neville Hilton (AFD Software) and Tim Drye (DataTalk)
kicked off the supplier presentations with a talk about their Censation
classification. Tim explained why segment titles can conceal rather
than reveal and how their product avoids this pitfall.
The second presentation, from Simon Perry (Beacon Dodsworth), discussed
Census issues and how the P2 People & Places classification is being
developed. Simon stressed the importance of testing classifications
in order to tell which works best and provided a methodology for comparing
test results.
In the last presentation of the morning, John Rae (CACI) explained
the two-stage approach taken in developing the new acorn classification.
John also discussed some of the internal requests that the developers
had turned down and why these decisions were taken.
After lunch, Martin Bradbury (EuroDirect) discussed the dangers of
taken an overly complex modelling approach to targeting, as opposed
to using a straightforward geodemographic discriminator. With the aid
of two case studies, Martin suggested that the discriminatory power
of the CAMEO classifier had increased, using 2001 Census data.
The next presenter, Rob Haslingden (Experian), discussed the importance
of understanding societal change in Britain. Rob talked about the impact
of using non-Census data to identify consumer segments and how this
will facilitate updating the UK mosaic classification.
The third presentation in this section was given by Eli Kling (GeoBusiness
Solutions). Eli explained the dimensions used in the ATOMIC product
and discussed the benefits of data mining using customer data linked
to the census.
After the break, Gordon Farquharson (Streetwise Analytics) discussed
the importance of developing spatial measures to overcome regional effects,
for example in house prices, and identifying segments of people most
likely to change in the future.
The final presentation of the day, from David Griffiths (TRAC Consultancy),
described the development of the Sonar classification. David explained
how a large number of alternative solutions were tested in order to
optimize performance.
The seminar ended with a lively panel session, in which many members
of the audience asked questions and made comments and observations.
This session included a heated discussion between the speakers on the
relative merits of area and household classifications.
Barry Leventhal
Chair,
MRS Census and Geodemographics Group
Back to Networking Introduction
|
Site Map
