44th Congress of the European Regional Science Association: "Regions and Fiscal Federalism", 25th - 29th August 2004, Porto, Portugal
One of the main questions to solve when analysing geographically added information consists of the design of territorial units adjusted to the objectives of the study. In fact, in those cases where territorial information is aggregated, ad-hoc criteria are usually applied as there are not regionalization methods flexible enough. Moreover, and without taking into account the aggregation method applied, there is an implicit risk that is known in the literature as Modifiable Areal Unit Problem (MAUP) (Openshaw, 1984). This problem is related with the high sensitivity of statistical and econometric results to different aggregations of geographical data, which can negatively affect the robustness of the analysis. In this paper, an optimization model is proposed with the aim of identifying homogenous territorial units related with the analyzed phenomena. This model seeks to reduce some disadvantages found in previous works about automated regionalisation tools. In particular, the model not only considers the characteristics of each element to group but also, the relationships among them, trying to avoid the MAUP. An algoritm, known as RASS (Regionalization Algorithm with Selective Search) it also proposed in order to obtain faster results from the model. The obtained results permit to affirm that the proposed methodology is able to identify a great variety of territorial configurations, taking into account the contiguity constraint among the different elements to be grouped.