THE ANALYSIS OF TIME SERIES FORECASTING METHODS, MODELS AND SOFTWARE PACKAGES
Анастасія Олегівна Долгіх, Олег Григорович Байбуз · Open Information and Computer Integrated Technologies · 2018
The analysis of existing for today’s day time series forecasting methods and models and their classification is carried out. The description of both classical statistical forecasting methods and models, such as exponential smoothing models, ARIMA models, regression models, and more modern ones based on the principle of machine learning, for example, the model based on support vectors (SVM), neural networks, markov models and classification-regression trees is presented. Software packages that contain modules of time series analysis and forecasting are described. The directions of future research and possible software developments in the field of financial time series forecasting are presented.