Clustering cities based on their development dynamics and variable neigborhood search

Abstract

Clustering cities based on their socio-economic development in long time period is an important issue and may be used in many ways, e.g., in strategic regional planning. In this paper we continue our recent study where cumulative attribute for each year replaces nine other attributes, called 'vector of dynamics'. In our previous paper some original ranking method was proposed. Using the same data set, here we try out some classical clustering models such as Minimum sum of squares and Harmonic means clustering. Results for the two last models are obtained using Variable neighborhood search based heuristics. A comparative study among old and new results on 120 Russian large cities are provided and analyzed.