Development of Predictive Models of Socio-Economic Systems Based on Decision Trees with Multivariate Response

–The work is devoted to the actual problem of constructing decision trees with a multidimensional response of the optimal structure, which are used to create predictive models for the evolution of complex systems. The aim of the work is to generalize the experience of constructing decision trees with a multidimensional response and to study the homogeneity and violation of symmetry of classes of models of socio-economic systems based on decision trees, which most clearly show the process of changing the states of the system and filling the space of possibilities, as well as signs of self-organization, which is cause of evolutionary processes and a consequence of symmetry breaking. An example of building a tree with a multidimensional response for a credit scoring problem is shown. The approaches described in the work show the connection between the phenomenon of symmetry breaking and the phenomenon of heteroscedasticity of regression models. The possibility of overcoming the problem of instability of finite predictions of models based on decision trees by developing approaches to the study of the heteroscedasticity of predictive models of socio-economic systems and the homogeneity of groups of objects is considered. Keywords––economy, information, clustering, regression, decision trees, heteroscedasticity, asymmetry