Year: 2026
Type: article
Abstract: This paper proposes an automated intervention analysis algorithm to assess the impact of external events on time series data.The proposed method automatically searches for optimal combinations of intervention functions without requiring prior specification by the analyst, thereby reducing subjectivity in model construction.The algorithm systematically evaluates candidate models and selects the best-fitting one based on commonly used evaluation criteria such as the Akaike Information Criterion (AIC), Bayesian Information Criterion (BIC), and Root Mean Square Error (RMSE).This automated procedure enables empirical and reproducible analysis while alleviating the need for expert knowledge of intervention modeling and time series theory.As a result, the proposed framework provides a user-friendly analytical environment even for non-experts unfamiliar with time series methods.To demonstrate its practical applicability, the algorithm is applied to three real-world time series datasets that exhibit structural changes caused by external events.The results are compared with those obtained from existing approaches, illustrating that the proposed method effectively captures diverse intervention effects and offers a scalable and objective alternative to conventional intervention analysis.
Authors: Leekyung Kim, Byeongchan Seong
Institution: Chung-Ang University
Language: English
Subfield: Signal Processing
Field: Computer Science
Domain: Physical Sciences
Open Access status: diamond
Funder: Chung-Ang University
OpenAlex ID: W7126208374