AUTOMATIC ALGORITHMS FOR ARIMA MODEL IDENTIFICATION IN TIME SERIES FORECASTING: A LITERATURE REVIEW
Authors
Keywords
Forecasting automation, ARIMA models, Automatic algorithms for
identification of ARIMA models.
Summary
Forecasting plays a crucial role in effective planning and management within both the
economic and public sectors, especially in a rapidly evolving environment. Despite its
importance, producing accurate forecasts requires considerable expertise and is a labourintensive
process prone to subjective errors, especially when processing large volumes of data.
These challenges make the role of automation of forecasting processes increasingly necessary
in modern practice.
The objective of this study is to derive the advantages and shortcomings of various
automatic algorithms for ARIMA model identification and to determine the extent to which
these algorithms reduce subjectivism in the modelling process.
The research includes three main tasks: (1) to justify the necessity of automation of the
forecasting process and reveal the challenges to its implementation, as well as to trace its
evolution; (2) to motivate the automation of ARIMA model identification on the basis of
revealing their nature and the key stages in the modelling process; (3) to present and compare
automatic approaches that solve ARIMA model identification challenges in order to further
evaluate their effectiveness and applicability.
The results of the study show that automated ARIMA algorithms provide high accuracy
and reliability of forecasts, significantly reducing the time and resources required for the
analysis. In addition, automation facilitates adaptation to changing data and conditions, which
is key to making timely decisions in a dynamic environment. The findings of this study can
assist organisations in selecting appropriate automated methods for time series forecasting.
Pages: 32
Price: 3 Points
