Application of methods of machine learning and data mining to problems of institutional economics

Speakers

Anna Kuznetsova (Institute of Biochemical Physics RAS)Mr Igor Kirilyuk (Institute of Economics RAS)

Description

Methods of machine learning and data mining are widely used now in many tasks of physics, chemistry, biology, medicine and in some other areas. Such techniques allow to obtain reliable results in poorly formalized areas. The last several years we used these methods for search of regularities in panel data of the statistical indicators, characterizing the institutional economy of different countries of the world. Consideration of influence of institutional characteristics of the countries connected with features of the legislation (for example, the property rights), traditions, habits of citizens of the concrete countries on indicators of economic development is important for borrowing of their experience, including, economic. Our analysis is based on variety of machine learning tools including some original techniques. For verification of the identified regularities the methods based on permutation tests, which do not demand aprioristic assumptions of the nature of probabilistic distributions, and allowing correct application at the small data set. Panel data researches usually are associated with analysis of a large number of the factors, characterizing relative dynamics of economic indicators. So they demand adequate accounting of multiple testing problem. This work was supported by RFBR (grant №17-02-00207)

Primary authors

Anna Kuznetsova (Institute of Biochemical Physics RAS) Mr Igor Kirilyuk (Institute of Economics RAS) Mr Oleg Senko (FRC Computer Science and Control of RAS)

Presentation materials

There are no materials yet.