Cluster-Based Demographic Typology of Rural and Suburban Municipalities: Case of Latvia

Aleksandrs Dahs , University of Latvia
Juris Krumins, University of Latvia

Contemporary goals of regional convergence are difficult to achieve, considering the uneven demographic scenery, competition between administrative-territorial units and limited resources available for targeted regional demographic policy measures. In the context of sustainable regional demographic development it is critically important to have an ability to quickly and effectively evaluate demographic risks and assets associated with specific regions or territories. In this study we aim to develop, test and evaluate methodology for an efficient demographic classification of rural administrative-territorial units using adaptive non-hierarchical clustering, while using the recently reformed municipalities of Latvia as an example. Data used for this purpose include statistical datasets of Latvian municipalities recently recalculated according to the new administrative-territorial division. Our proposed methodology relies on a robust k-medoids PAM (Partitioning around Medoids) clustering algorithm, as it is not reliant on a pre-defined measures of distance or dissimilarity between observations and allows for greater freedom in number of indicators and types of data used. Study results show that with carefully selected data sets, a non-hierarchical cluster analysis employing k-medoids or similar method can become a feasible tool for quick and efficient classification of territorial units according to their socio-economic and demographic characteristics. The established clustering technique provided most useful results when carried out in several iterations, using different data cross-sections. We can also conclude that the demonstrated approach permits application of different grouping criteria sets, thus giving room for experimentation and approbation of various theoretical frameworks.

See extended abstract

 Presented in Session P1. Postercafe