ARTICLE
ABSTRACT
The general idea of statistical tests which allow testing the influence of geographical or administrative units of upper level on clustering results of lower level units is presented, basing on the authors' earlier works. The so-called “active border” notion is used in these methods. If two counties (powiats) have been classified into different clusters then the border between them is called active. This border can be also the border between provinces. The number and length of active borders are used in the proposed test statistics. Their distribution depends on the actual geographic division of a given country. In this paper we present results for Poland and division for provinces and counties. Tables for test critical values and the approximation functions are given
KEYWORDS
cluster analysis, NUTS, comparing partitions.
REFERENCES
SOKOŁOWSKI, A., (1979). Generowanie losowego podziału zbioru skończonego. Prace Naukowe Akademii Ekonomicznej we Wrocławiu [Generating random distribution of a finite set. Scientific Papers of Wrocław University of Economics], 1979, 160 (182), 413–415.
SOKOŁOWSKI, A., STRAHL, D., MARKOWSKA, M., SOBOLEWSKI, M.,(2013a). The influence of upper level NUTS on lower level classification of EU regions. European Conference on Data Analysis, Luxembourg, July 10–12, 2013.
SOKOŁOWSKI, A., STRAHL, D., MARKOWSKA, M., SOBOLEWSKI, M., (2013b). The hierarchy test of geographic units based on border lengths, Conference of the International Federation of Classification Societies IFCS 2013, Tilburg, Netherlands, July 14–17, 2013. Abstract in Program and Book of Abstracts, Tilburg University, 46
