Forecasting Stock Quotes Using Multidimensional Classification Methods

This work examines the theory of predicting cryptocurrency exchange rates, the implementation of various methods and the development of the service. The purpose of the work is to develop a service for a hybrid model of forecasting the cryptocurrency exchange rate. The object of research of this final qualifying work is a hybrid model of multidimensional multi-stage stock price forecasting. The subject of the study is the architecture of a hybrid model of multidimensional multi-stage stock price forecasting. The final qualifying work is divided into the following chapters: introduction, three main chapters, conclusion, list of sources and appendices. The first chapter includes theory: research on forecasting stock prices and cryptocurrencies, description of the features of the cryptocurrency market, conclusion and formulation of the problem. The second chapter describes the methods. It discusses the main methods used in solving time series forecasting problems, as well as describes the principle of operation of hybrid machine learning models. The third chapter describes the development and comparison of hybrid models and the implementation of the service for the most effective model. The volume of work is 57 pages. The work contains 14 figures, 2 tables and 6 appendices, 32 sources of information were used.