Skip to Main content Skip to Navigation
Conference papers

Les indicateurs locaux d'association spatiale (LISA) comme méthode de régionalisation : Une application en Inde.

Abstract : In India, quality of census data is good enough to allow geographic approach at micro-scale (village level) of the socioeconomic phenomena. However, variability of data at this scale, as well as the huge number of units to take into consideration, make it difficult to present synthetic results, which would underline the major trends of the space structure. Our aim is to present a solution, which, following Gould's expression, let “the data speak from themselves”. The goal is to show the regional trend while keeping the local details visible. For this purpose, we use the local indicators of spatial association (LISA), which let us measure similarities (and dissimilarities) between neighbours. LISA have been developed by Luc Anselin (1995) and allow us to compute the level of likeness of one spatial unit with its neighbours (with the help of a weight matrix based on contiguity, distance or nearest neighbours). We are then able to show regional trends without losing local values, i.e. to keep the information relative to the internal heterogeneity of the homogeneous zones. Our paper will introduce the statistical dimension of LISA in order to understand how they run and their limits. We will then propose an application of LISA to create an original division of Tamil space, based on homogenous regions. This state, located in the South of India, has more than 60 millions inhabitants (in more than 16.000 villages), which constitutes a good case study to test the efficiency of this tool.
Document type :
Conference papers
Complete list of metadatas

Cited literature [2 references]  Display  Hide  Download
Contributor : Sébastien Oliveau <>
Submitted on : Sunday, September 19, 2010 - 4:11:38 PM
Last modification on : Thursday, July 2, 2020 - 7:53:31 AM
Long-term archiving on: : Monday, December 20, 2010 - 2:45:05 AM


Files produced by the author(s)


  • HAL Id : halshs-00519265, version 1


Sébastien Oliveau. Les indicateurs locaux d'association spatiale (LISA) comme méthode de régionalisation : Une application en Inde.. 7ème rencontres ThéoQuant, Jan 2005, Besançon, France. pp.1-10. ⟨halshs-00519265⟩



Record views


Files downloads