Geolocalized data in France Benefits from the GeoStat project Geostat++(FR) Jean-Luc LIPATZ INSEE - France October 2010

Transcription

1 Geolocalized data in France Benefits from the GeoStat project Geostat++(FR) Jean-Luc LIPATZ INSEE - France

2 France is different An unusual status for the French NSI Data provider and disseminator, but also : Making analyses by itself Making analyses in close cooperation with local authorities Experience using gridded data since 2006 The National Committee for Statistical Information (CNIS) September 2009 : publication of a report from a task force about the spatial distribution of population, animated by the director of the town planning agency for Bordeaux. Explicit request for dissemination of gridded data, on a wider geographical range than that is currently available Page 2

3 Troisième point de convergence : la nécessité de faire évoluer la géographie de production et de diffusion des statistiques territoriales Un fichier fournissant des estimations de population et d emploi (et si possible de personnes en formation ) selon un maillage «carreau» de surface à définir et/ou selon un maillage parcelle cadastrale avec la surface (pour calculer des densités). Page 3 Rapport du groupe de travail STATISTIQUES ET NOUVELLES TENDANCES DE LOCALISATION DES POPULATIONS ET DES ACTIVITES SUR LE TERRITOIRE Président : Francis CUILLIER

4 gridded data where we can What happens there? Page 4

5 General context The two main obstacles Administrative registers cannot be matched together on the basis of an individual identification number There is no general purpose address register to reference to Insee has built its own with limited objectives The NMA has only a partial one Some parts of the territory have very fuzzy postal addresses Page 5

6 Agenda The master : Census The slaves : Administrative sources Liberation! Examples Page 6

7 The French census Data collection takes place each year on a sample, results are elaborated using five years of data collection Municipalities above inhabitants Based on a sample from a reference address register 8 % of the addresses each year Address register is maintained within a GIS by Insee it self, home made geometry available Municipalities below inhabitants Classical traditional data collection 20 % of the municipalities each year Geometry of census districts is not available Page 7

8 Gridded data from census? No plan to change the data collection process No plan to make standard (=native) estimations Not easy where it can be done (municipalities above inh. : 50 % of the whole population, 4 % of the territory):grids with large size of the cells can use the same estimation process than the one used for the sub municipality standard output areas Step 1) Estimation with the 5 x 8 % sample Step 2) Correction of the estimate by fitting the number of dwellings with the one known in the address register Grids with small size of the cells can use a mixed estimation system, using an auxiliary data source See 2007 EFGS conference Available gridded data will be driven by explicit user needs Page 8

9 Administrative data Administrative data can only be geolocalized by interpretation of postal addresses + Today uses the census reference base of addresses (over inh.) + In the future, could get benefits from a French project to build a reference base of postal addresses for the whole France But conversion from addresses to coordinates is too expensive to be done on a wide range : manual processing is required for about 10 % of the addresses A part (5 %) of the addresses cannot get precise coordinates, estimation takes place gridded data is easier to compute (no sample), but getting coordinates for making takes time and resources Page 9

10 Administrative data (2) Fiscal data Fiscal data reference the cadastral register : Cadastral parcels can be searched for within maps displayed on a web site The tax for housing register references cadastral parcels The income tax register references the tax for housing register An agreement with the cadastral services was signed beginning 2010, whole cadastral register was delivered at the end of june. Geostat project was one of arguments for the agreement The other argument was the delineation the urban zones : the actual process needs estimations of population in the built up areas within the municipalities : such estimations are a by-product from the Geostat project. Page 10

11 gridded data in France Coordinates but not everywhere? Revolution is on the way Fiscal data (income and housing taxes) + Now (June 2010) linkable to the cadastral register in an automatic (=cheap & quick) way to get location of dwellings + Fiscal data is rich enough to provide a basis for estimating census like data - Fiscal population is biased (week end time population?) - Cadastral register is not available in a convenient format for 1/3 of the municipalities + The available 2/3 cover the needs for urban analysis + Census address register is no more needed! Page 11

12 173 rue LESAGE What is an address? IRIS 2601 The building The center of the property unit The address for census data collection IRIS 2503 Page m grid

13 Geolocalized data The new geographical extent Population from fiscal data, Aggregated into 1km² grid cells Page 13

14 After having filled the gaps Page 14

15 1km 25 km 200m 100m Page 15 JL.LIPATZ EFGS meeting

16 A city in its environment Page 16

17 But analysis will require more a detailed view Page m grid

18 that was not available before Page 18

19 Extended opportunities for analysis Simplified population density, 200 m wide grid cells Removing noise fˆ( x) N i1 1 x X K Nh h i Page 19

20 Extended opportunities Risk estimate for being in the 1 st quartile of income Removing noise, twice fˆ( x) N i1 N i1 1 xyi K Nh h 1 x X K Nh h i Page 20 Officially deprived neighborhoods (1 st generation)

21 Extended possibilities of diagnostic Automatic delineation of areas of interest Red boundaries : automatic selection Blue boundaries : deprived areas as currently defined 44 % 38 % 47 % 56 % Officially deprived neighborhoods (2 nd generation) 46 % 27 % 35 % 42 % Page % City mean : 1/4

22 Conclusion? Households with reference person above 65 Using grid level data Using municipalities level data Page 22

23 Conclusion! the surroundings of Dijon Thank you Any questions? Before After Next step : Disseminate! Page 23