A method of data mining on the state of industrial sectors in the regions of Russia, taking into account social factors, was developed. The method is based on the development of a hierarchical structure of initial features and application of the principal component analysis according to the constructed structure. The necessity of decomposing a set of the initial features is justified, due to the fact that the results of the preliminary principal component of a large number of features turned out to be cumbersome and difficult to interpret. A two-level decomposition tree of the set of the initial features was developed. The analysis is carried out in accordance with the developed tree structure of the initial features. Three samples of the lower level and one sample of the upper level were selected. The use of the principal component analysis of the lower-level samples allowed us to assess the state of regions from various viewpoints: the development of the extractive industry, manufacturing industry and electric power industry, taking into account social factors. Fuzzy rules for clustering the regions in the space of the features of the lower-level samples are formed. Based on the constructed principal components for the samples of the lower level, an integral sample of the upper level is formed, which allows assessing the level of development of the regions as a whole. The proposed method of data mining should be used to compile a database of the fuzzy rules of an intelligent decision support system for managing the socio-economic development of the regions.
due to the urgent need to ensure economic growth in all sectors of the economy. The government is
developing concepts and strategies aimed at ensuring sustainable development of the state. The
Strategy for the Spatial Development of the Russian Federation for the period up to 2025 emphasizes
the importance both of the formation and strengthening of mineral resource centers, and development
of industries in the branches of promising economic specializations [
A number of works are known to be related to the construction of regional clusters and analysis of
regional differences, for example, both in terms of the level of development of the real sector [
of the Russian Federation for 2019 [
mining, the set of the initial features is decomposed into three directions which determine the samples
of the initial features of the lower level (see Ошибка! Источник ссылки не найден.). The choice
of a set of the initial features is determined both by the purpose of the study being carried out and by
the composition of the data available for the analysis [
The decomposition tree is built, firstly, based on the purpose of the study and, secondly, on the results of the preliminary studies. According to the purpose of the study, the first two directions are highlighted, associated with the construction of clusters of the regions which differ in the level of economic development, as well as the third direction which takes into account regional differences in the level of social development. Based on the results of the preliminary studies by the method of principal components, the possibility of integrating the features characterizing a particular industry sector into a separate component is determined. The features characterizing the branches of MI and
The features that characterize the level of socio-economic development of regions of Russia The features that characterize the level of economic development of the regions in the EI, MI, EP aspect The features that characterize the level of economic development of the regions in the MI, EP aspect
form an integral sample, which characterizes the level of development of the industry of the regions as a whole, taking into account social factors. The formed integral sample is also subjected to the data mining procedure. The results of the data mining analysis of the lower-level samples and integral sample at the upper level are the clusters of the regions and their characteristics, which are refined when moving up the tree of the initial features (see Ошибка! Источник ссылки не найден.).
level of economic development of the regions in the MI, EP aspect
For the first sample of the lower level of decomposition (Ошибка! Источник ссылки не найден.), using the method of principal components, the level of development of enterprises in the branches of EI, MI and EP was analyzed in the regions. The results of the analysis showed that in the first two principal components (PC), more than 91% of the variance was achieved. The weight coefficients of the features involved in the names of the principal components are presented in Ошибка! Источник ссылки не найден. and highlighted in bold type. The application of varimax rotation confirmed the composition of the identified principal components.
The weight coefficients of the features obtained as a result of the analysis show that: PC 1 allows identifying the regions with the high and low level of economic development of MI and EP, taking into account the number of enterprises and organizations of EI; PC 2 allows dividing the regions according to the level of economic development of EI. When conducting the component analysis, it is revealed that the most developed region in terms of the level of economic development of MI and EP is Moscow, and the most developed one in terms of economic development of EI is the Tyumen region. The leading regions stand out significantly in the scatter diagram and do not allow the visual assessment of the location of the remaining regions; therefore, the leading regions were removed from the initial data sample. The component analysis was repeated, and the composition of the main components did not change. The scatter diagram obtained during the analysis without taking into account the leading regions is shown in Ошибка! Источник ссылки не найден..
cluster analysis was made (Ward's method, Squared Euclidean metric), which, in general, confirmed the composition of the clusters identified during the component analysis of the data.
city of St. Petersburg, and the Moscow region are distinguished by the high level of development of
characterized by the high level of development of EI. A number of regions, for example, the Republic of Tatarstan, and Krasnoyarsk Region, are in the area of the balanced development of industries.
example, a rule for cluster 1 is given: IF the volume of goods shipped MI = small AND the number of enterprises and organizations MI = small AND the turnover of organizations MI = small AND the volume of goods shipped EP = small AND the number of enterprises and organizations EP = small AND the turnover of organizations EP = small AND the volume of goods shipped EI = small AND the number of enterprises and organizations EI = small, THEN Cluster = 1.
The analysis of the first sample of the lower level shows that the level of development of enterprises of MI and EP is separated into a separate principal component, while it is impossible to divide the regions with the predominant development of enterprises of either MI or EP. Therefore, the second sample of the lower level of decomposition was formed (Ошибка! Источник ссылки не найден.). The analysis of the level of development of the regions in terms of MI and EP was made in order to single out individual PCs for MI and EP. The analysis by the method of principal components shows that in the first three main components, more than 94% of the variance is achieved. The resulting weights of the signs (see Ошибка! Источник ссылки не найден.) allow us to conclude that: PC 1 characterizes the level of development of MI and EP in the region; PC 2 allows identifying the regions with the priority development of either MI or EP; PC 3 makes it possible to identify the regions in which both MI and EP are developed.
Volume of goods shipped (MI)
Volume of goods shipped (EP) Share of MI in gross value added
Share of EP in gross value added Number of enterprises and organizations (MI) Number of enterprises and organizations (EP) -1,5 2 C P -3,5 -5,5 -7,5 Turnover of organizations (MI) Turnover of organizations (EP)
makes it possible to form the following regularities. The city of Moscow is distinguished by the greatest development of MI and EP. In the Lipetsk and Kaluga regions, the strongest predominance of
Lipetsk region
Republic of Bashkortostan
Sverdlovsk region Republic of Tatarstan
Moscow region
Saint Petersburg Tyumen region Moscow
Chukotka Autonomous District -2 2 6 10 14 18 PC 1
найден.), the analysis of social factors was made. Based on the weights obtained for the signs, the following conclusion was drawn: the first PC characterizes the material well-being of the population, taking into account the level of employment in the region; the second PC characterizes the level of unemployment and the coefficient of tension in the labor market (with negative signs). These results are presented in the scatter diagram (Ошибка! Источник ссылки не найден.).
Saint Petersburg
Moscow Tyumen region The Republic of Dagestan Magadan Region
weight coefficients of the features (see Ошибка! Источник ссылки не найден.) indicate that: PC
The level of economic development of MI and EP, taking into account the number of enterprises and organizations of EI The level of economic development of EI The level of economic development of MI and EP The regions with the priority development of either MI, or EP The regions in which both MI and EP are developed Material well-being of the population, taking into account the level of employment in the region The unemployment rate and labor market tension coefficient
following regularities. During the component analysis of the integral sample (sample of the top-level) along PC 1, the regions are distributed according to the structural features of the development of MI and EP in the region. The most developed region is Moscow, and the Tyumen region, Moscow region,
distributed according to the level of social well-being in the region, and the highest values of this indicator are characteristic of such regions as Moscow, the Moscow region, the Sverdlovsk region, the
2C -1,9 P 2,1 0,1 -3,9 -5,9 Sverdlovsk region
Moscow region Republic of Tatarstan
Saint Petersburg Magadan Region Tyumen region Chukotka Autonomous District Moscow -2 1 4 10 13 16
A method for data mining is proposed, according to which a set of the initial features is decomposed to analyze the level of industrial development in Russian regions from various viewpoints: the development of the extracting industry, manufacturing industry and electric power industry, taking into account social factors. The composition of the formed principal components is used to form a sample of the top-level and its analysis, which makes it possible to assess the level of industrial development in the regions as a whole. It is advisable to use the proposed method of data mining to compile a base of fuzzy rules for an intelligent decision support system in managing the socio-economic development of the regions.
This work was supported by the RFBR (The Russian Foundation for Basic Research), grant No 20-08-00796 "Intellectual control of an industrial complex as a dynamic multi-agent object based on cognitive modeling and machine learning methods". 7. References