June 26, 2012, 14:00–15:30
Toulouse
Room MF 323
Statistics Seminar
Abstract
The analysis of socio-economic data often involves combining data bases from several administrative sources collected on different spatial partitions of the area of interest. It is then necessary to transform the data from the original spatial units (“sources”) to the final spatial support (“target”). One possible issue to this problem is to consider a regular grid (“carroyage” in French) and to transfer all data bases from their original support to this unique grid. This option is currently under study at INSEE and in Europe with the EU Inspire directive (INfrastructure for SPatial InfoRmation). We survey the literature on this topic called areal interpolation. There are three main classes of methods: dasymetric techniques, regression techniques and smoothing techniques. They differ by the type of target variable (continuous or discrete, intensive or extensive), the satisfaction or not of the picnophylactic property, and other simplifying assumptions. ............... CE SEMINAIRE S’INSCRIT EGALEMENT DANS LE CYCLE D’EXPOSES DU GROUPE D’ECONOMETRIE SPATIALE