Title: Delineating sub city districts for decision making An attempt to automate the process
1Delineating sub city districts for decision
makingAn attempt to automate the process
2Context
- Urban policies are heavily territory based
- 751 Zones urbaines sensibles (ZUS)
- Need for an evaluation
- Zones in use where defined in 1996-1997
- Town planning has modified the structure of
several urban areas - Urban riots of end of 2005 needed a political
response (within suburbs, millions are set on
fire too) - Need for extensions
- New zones are already in use locally (800)
- Formalized additions already exist for specific
purposes
3Question (December 2005)
- For the national level (urban affairs ministry)
- Target cities and zones absolutely needing
attention - Give a filter to set priorities among locally
defined zones - For the local level (regional government and
cities) - Provide data to quantify problems in selected
zones - Provide methodological help to find or to argue
for new zones
4Challenge no 1Find a relevant basic block
- Standard geographical units are not relevant
- Block size around 2 000 inhabitants, existing
zones ranges from 500 to 50 000, target for new
one has only a (theoretical) lower bound of 2
000. - Design essentially based on form of the built up
area, not suitable for measuring anything else - gt Design zones without using existing zoned data
5The trap with standard geographical blocksfrom a
classroom example
Standard zones for dissemination
1000 inhabitants Unemployment 10
1000 inhabitants Unemployment 10
Unobservable zones
1000 inhabitants Unemployment 20
1000 inhabitants Unemployment 20
Deprived neighbourhood
6 to real cases Unemployment in City of
Gennevilliers,
Available on the web,using standard
dissemination blocks
From individual data,using spatial analysis
tools
7Not only a problem when drawing mapsUrban
segregation ranking among large citiesaccording
to Duncan and Duncan index(Low/high
qualification households)
Basic geographical block size (standard
dissemination blocks)
2 000 inhabitants
5 000 inhabitants
8Challenge no 2Find relevant sources
- Recent, available at detailed geographical level
within a short timing? - Census ok but too old (1999)
- Administrative sources, recent but
- Disconnected themes income, unemployment,
employment - Biased no disposable income, no LFS
unemployment, etc - Covering large but not complete and different
fractions of the population - gt Each one must be used separately
- Geography partially processed
- gt To be finished with estimations of individual
location - gt Use rounded coordinates grided data (100m x
100m) - gt Prefer biased recent data to old but correct
data - Not THE answer but a set of disconnected answers
9The proposal a set of mapsshowing the rough
outlines of the zones having a high proportion
of deprived population, relatively to local
situation
Blue the zones currently in use (ZUS) Red
suggested areas where to design new zones Grey
spatial distribution of the global population
10Challenge no 3Human factors
- Inside the INSEE two revolutions
- Whole environment is different new data, new
tools - gttargeted training,
- Implies active involvement in a decisional
process - Outside finish the work!
- The proposal from INSEE is only a set of hints
for knowing where to create a new zone, not its
outline - Other (non quantitative) data must be used
local mood - Priority ranking has to be done
11From data to final map
1) Simplify the maps
Rough grid data to smoothed data
4)Superpose the maps
3) Extract the outlines
2) Combine the maps
12Simplifying the viewsNonparametric density
estimation
Longitudinal section, 100m step.
13Selecting the areasRelative thresholds
Selected areas
Ratio
Value of the threshold select the upper half
of inhabited zones more deprived than the whole
city average
14It works!Same data source, several cities(low
qualifications from wage payment registers)
Rennes
Rouen
North of Paris
Roubaix
15The need for several points of viewSame city,
several data sources La Rochelle
Administrative artefact
Low income from social family benefits
Low income from income tax registers
Already known as a potential zone
New zone
Low qualifications from unemployment registers
Low income from health insurance registers
16That is not science fiction!
- At national level
- One full set of maps was disseminated in the web
site of the urban affairs ministry (mid 2006) - Thresholds were used as a segregation index to
set priorities among cities - Dissemination of such maps within insee.fr is
planned for 2008 - At regional level
- Collaborative work initiated between regional
statisticians and many cities - Turned to be a more general analysis tool still
in use - Grided data production is no more an exception
17Thank youQuestions?