Kozhukhivska O. Models, methods and information technology for forecasting of nonstationary financial processes and associated risks

The work is dedicated to solving the actual scientific and applied problem of development of models, methods and information technology of forecasting the non-stationary financial processes and associated risks in conditions of uncertainty of financial and economical processes development. There have been developed a method of building the hybrid forecasting model on the basis of a combination of the method of group consideration of arguments and neural networks, a method of structural and parametric synthesis of nonlinear non-stationary processes of models based on hybrid structures of MGUA-Neural networks, optimum Calman filter for non-stationary nonlinear (heteroscedastic) processes. Using the proposed new and improved existing models and methods of heteroskedastic processes, there has been developed a software-based information system for decision making support, characterised by versatility and expansion of its Functional capabilities for solving practical modeling problems, forecasting the development of non-stationary processes and assessing the associated financial risks. The results of the work have been introduced in banking institutions and insurance companies. The results of the dissertation work are introduced in the form of methods and means of mathematical models constructing nonlinear non-stationary processes in the educational process of Department of Quality, standardization and project management of Cherkasy National University named after Bohdan Khmelnytskyi.