A Review of Machine Learning Methods for Long-Term Time Series Prediction
Milan P. Ptotic, Miloš B. Stojanović, Predrag M. Popović · 2022 57th International Scientific Conference on Information, Communication and Energy Systems and Technologies (ICEST) · 2022
This paper analyses some of the fundamental machine learning methods for time series predictions. Standard prediction methods, such as the Box-Jenkins ARIMA approach, are discussed, as well as the supervised machine learning methods where the focus is on the artificial neural network and the support vector machine. Overall, five strategies used for long-term prediction are compared and some aspects of future research are discussed.