The emerging structure of Russian urban systems: a classification based on Self-Organizing Maps

This paper argues that there is a complex socio-economic, spatial and political trend towards increasing unevenness among Russian cities against the shortage of researches about Russian urbanized space. In addition to this Russia experiences a lack of studies taking into account that, in modern globalized world, cities are considered as a power machine stimulating the country's development. The urban researches on Russian cities mainly are focusing on Moscow and Saint Petersburg, or on some regional capitals and cities hosting mega-events such as Vladivostok and Sochi. As a matter of fact, a considerable part of Russian cities have lost themselves in the new market conditions since the general vector of urban development has changed after the USSR breakup. Great transformations have occurred in both inter and intra urban levels. The disparity of country's development has increased: European Russia and Northern regions with rich oil and gas deposits are becoming more affluent, while Far East and Siberian regions are undergoing population loss and resources outflows. In spite of contemporary Russian policy is mainly focused on the national and regional issues, world economy is more and more aware that cities are the growth poles for a whole country. Therefore an updated development policy demands a re-scaling at the urban level and requires a precise analysis of urban condition and dynamics. The paper aims to classify the whole Russian urban system on the base of some socio-economic characteristics: demographic dynamics, housing quality, economic performance at two temporal thresholds. The adopted method, the Neural Network Self-Organizing Maps (SOM), is able to single out groups of cities with high internal resemblance. The paper starts with a brief overview of the urban networks formation during soviet period, its transformation after the breakup of the USSR and the consequences of these two processes for the contemporary cities. The second section explains the data which will be used for the analysis and describes the used SOM algorithm. The subsequent section presents the analysis of the results describing spatial urban patterns in terms of quantities and geographical characteristics. The conclusions discuss the nature of those patterns. Due to SOM implementation it has been possible to identify twenty five groups of cities, with similar socio-economic trends during the last decade, where each group is characterized by an appropriate profile (a codebook). Moreover the empirical results have allowed sketching a new urban hierarchy in Russia, outlining four layers: 'urban engine', 'strong cities'; 'potential cities' and 'weak cities'. The outcomes will allow the definition of appropriate urban development strategies.