Geospatial Analysis of Consumer Data

Paul Longley and colleagues, CDRC and University College London

Groups

A1: Struggling suburbs

A2: Suburban localities

B1: Disadvantaged diaspora

B2: Bangladeshi enclaves

B3: Students and minority mix

C1: Asian owner occupiers

C2: Transport service workers

C3: East End Asians

C4: Elderly Asians

D1: Educational advantage

D2: City central

E1: City and student fringe

E2: Graduation occupation

F1: City enclaves

F2: Affluent suburbs

G1: Affordable transitions

G2: Public sector and service

employees

H1: Detached retirement

H2: Not quite Home Counties

A: Intermediate Lifestyles

E00007298; Perry Mead, Enfield

Groups

A1: Struggling suburbs

A2: Suburban localities

• Later stages in life-cycle

• White and born in the UK

• Few dependent children

• Most live in single family terraced

or semi detached properties.

• Higher social rented.

• Average employment in full and

part time intermediate

occupations.

• Lower levels of highest

qualifications

B: High Density and High Rise Flats

E00009768; Lancaster Court, Fulham

Groups

B1 Disadvantaged diaspora

B2 Bangladeshi enclaves

• Densely populated areas of flats.

• Families have children of school age

• Many residents Bangladeshi origins

• High Black residents or Mixed or

Other ethnic groups.

• Higher spoken language is not

English.

• Qualifications are below the London

average

• Some residents are full-time students

living in shared accommodation.

• Levels of unemployment and part-

time working high

• Employment more typically in

administration, or in accommodation

and food services industries.

C: Settled Asians

E00013190; Catherine Gardens, Hounslow

Groups

C1 Asian owner occupiers

C2 Transport service workers

C3 East End Asians

C4 Elderly Asians

• Traditional single-family

houses

• Above average numbers of

which are owner-occupied.

• Full age range

• Main language spoken in

many households is not

English.

• Occupations drawn from a

wide range of non-

professional sectors. Many of

Asian origins, although many

are second or subsequent

generation British residents.

D: Urban Elites

E00009327; Stoner Road, West Kensington

Groups

D1 Educational advantage

D2 City central

• Young professionals

• Working in the science,

technology, finance and

insurance sectors. Large

numbers of students

• Many privately owned flats

• Residents are

disproportionately drawn

from pre 2001 EU countries,

• High of Chinese, Arab and

other minority backgrounds.

E: City Vibe

E00009228; Netherwood Road, Shepherds Bush

Groups

E1 City and student fringe

E2 Graduation occupation

• Many young, single professionals

• Mostly living in Zone 2

• Few individuals originate from the

Indian sub-continent

• Mixed ethnic groups are well

represented, as are migrants from

pre 2001 EU countries.

• Large number student

households

• Individuals rent within the private

sector

• Well qualified

• Employed in a range of

professional, scientific and

technical occupations.

F: London Life-Cycle

E00017504; Trinity Road, Wimbledon

Groups

F1 City enclaves

F2 Affluent suburbs

• Predominantly White in ethnic

composition (including

individuals from other pre 2001

EU countries)

• Households cover the full

family life-cycle

• Residents are highly qualified

• Employment rates are high

• Employment is concentrated in

the technical, scientific,

finance, insurance and real

estate industries.

G: Multi-Ethnic Suburbs

E00022299; Crescent Road, Leyton

Groups

G1 Affordable transitions

G2 Public sector and service employees

• Wide range of non-White

ethnic groups

• EU post 2001 are well

represented.

• Young children or children of

school age,

• Low over 65s

• Family housing in

overcrowded terraces,

• social housing sector.

• unemployment are high.

• Employment blue collar

occupations.

H: Ageing City Fringe

E00005585; Peacock Gardens, South Croydon

Groups

H1 Detached retirement

H2 Not quite Home Counties

• Many residents 45+

• Many above state pensionable

age.

• High levels of marriage

• Mainly white

• Much of the dwelling stock semi-

detached and detached houses

• Levels of qualifications are low

• Private vehicle ownership is

high

• Levels of unemployment are

very low and drawn from a

range of sectors

London Output Area Classification

2011 Output Area Classification

http://www.google.co.uk/intl/en_uk/earth/

52: POORER FAMILIES,

MANY CHILDREN,

TERRACED HOUSING

51: YOUNG PEOPLE IN SMALL, LOW COST TERRACES

59: DEPRIVED AREAS AND HIGH-RISE FLATS

11: SETTLED SUBURBIA, OLDER PEOPLE

Urban Adversity

Affluent Achievers

Sponsors (2012-16)

Research Dissertation ProgrammeOutputs from 2015

www.cdrc.ac.uk/retail-masters/2015-projects/

CDRC Maps Geodemographic

Classification: OAC

CDRC Maps Modal Dwelling

Age Group

Internet User Classification

Towards the ‘Smart Census’

• Context of better use of existing data resources, e.g. workplace statistics

• Activity patterns associated with consumption

• Big Data as ‘exhaust’ (Harford): no research ‘design’

• ‘Horses for courses’ approach to data creation, maintenance and linkage

Data available through the Twitter API

• User Creation Date

• Followers

• Friends

• User ID

• Language

• Location

• Name

• Screen Name

• Time Zone

• Geo Enabled

• Latitude

• Longitude

• Tweet date and time

• Tweet text

Twitter estimated footfall in Soho

Time (hours) Time (hours)Time (hours)Time (hours)

Fre

qu

en

cy

Fre

qu

en

cy

Fre

qu

en

cy

Fre

qu

en

cy

The average weekday activity in 2013

• The frequency of

geotagged Tweets

across space and

time can tell us a lot

about the dynamics

of a city

Forenames – Age (Males)

5 clusters of forenames based on their

age distributions

Inferred demographic structure of Tweeters

The O2 Arena The Emirates stadium Canary Wharf Westfield Stratford

Twitter vs the Census (courtesy: Guy Lansley)

Day NightTwitter

Census

• Lower Super Output Area level

• Census work day statistics vs Tweets from 10:00 – 16:00

• Census residential population vs Tweets from 19:00 – 7:00

Customers most frequently visited store outside MSOA of residence[Courtesy: Alyson Lloyd)

Smart Meter Data4 clusters of smart meters based on typical daily energy profiles

From: Samson, N., Lansley, G. and Simpson, A. (2014) Using smart meter data to determine energy efficiency of customers’ homes. PopFest 2014. 4th – 6th August. University College London, UK.

Smart Sensors

• Co-production of Big Data

total

67.5k businesses

17.5m residents

total

17.5k businesses

3.4m residents

Size of the data

• 1000 sensors generate approximately 1.5 GB of data every day. This equals approximately 5-10m records every day.

• Compared to this, tracking 100,000 properties with 50 parameters updated every month generate 2 GB of data in 10 years.

100,0

00 p

roperty

data

base

1000 s

ensor data

base

Comparison of data generated by the sensors to a retail property database in a year

Complexity of visualisation

• Footfall counts

• Trends

• Hourly footfall

• Hourly trends

• Relationship between the sensors

01 Jan 2015, Fri

05 Jan 2015, Tue

04 Jan 2015, Mon

03 Jan 2015, Sun

02 Jan 2015, Sat

* from sensors at Market Harborough

Classification

Classification

Classification

Some prospects

44

Clapham Junction

Victoria

Waterloo

London Bridge

Liverpool Street

Fenchurch Street

St Pancras

Kings Cross

Euston

Paddington

Marylebone

Lewisham

Topic 6 Subgroup B – Trains and Delays

Underrepresented Overrepresented

Topic 13 Subgroup D – Education

UCL

University of

Westminster

Imperial College

LondonLondon South

Bank University

Kings College

London

Queen Mary

London Metropolitan

University

University of

Greenwich

City University

Goldsmiths

Birkbeck

SOAS

LSE

Various

Various

Underrepresented Overrepresented

Some prospects• Rethinking ‘place’ as the

measurable accumulated effects of slow and fast dynamics

Courtesy:

James

Cheshire

Isonymy groups (left) and the geographic distribution of genotypes (right)

[courtesy Jens Kandt]

Socio-economics• There is an association between names and socio-economics

and geodemographics

• E.g. Top 5 forenames for each 2011 OAC Supergroup

Rural Residents Cosmopolitans Ethnicity CentralMulticultural

Metropolitans

PENELOPE TOM MOHAMED MOHAMMED

HUGH NICK AHMED MUHAMMAD

ALASTAIR HARRIET ALI MOHAMMAD

ROSEMARY MAX JOSE ABDUL

PHILIPPA ALEX ABDUL AHMED

Urbanites SuburbanitesConstrained City

Dwellers

Hard-Pressed

Living

TOBY HILARY LILLIAN KAYLEIGH

PHILIPPA GEOFFREY MAY LEANNE

JEREMY KATHRYN ETHEL LYNDSEY

KATHERINE JILL KAYLEIGH STACEY

DUNCAN GILLIAN ELSIE KYLE

Data: 2011 Enhanced Electoral Roll (CACI UK Ltd)

Holborn10:30 – 11:10

Holborn Underground station pace

DemographicsCommute

Energy. . .

Some prospects• Rethinking ‘place’ as the

measurable accumulated effects of slow and fast dynamics

• New ways of framing research questions, e.g. segregation, health outcomes

The local geodemography of Glasgow, showing the 7.8 mile route that links

communities with life expectancies of 54 and 82 [courtesy Alex Singleton]

Thank you.

p.longley@ucl.ac.uk