Bayesian optimization based dynamic ensemble for time series forecasting
Among various time series (TS) forecasting methods, ensemble forecast is extensively acknowledged as a promising ensemble approach achieving great success in research and industry. Due to the high diversification of individual model assumptions, heterogeneous information fusion contributes to genera...
| Main Authors: | , , , |
|---|---|
| Other Authors: | |
| Format: | Journal Article |
| Language: | English |
| Published: |
2022
|
| Subjects: | |
| Online Access: | https://hdl.handle.net/10356/163881 |
| _version_ | 1826117573487362048 |
|---|---|
| author | Du, Liang Gao, Ruobin Suganthan, Ponnuthurai Nagaratnam Wang, David Zhi Wei |
| author2 | School of Civil and Environmental Engineering |
| author_facet | School of Civil and Environmental Engineering Du, Liang Gao, Ruobin Suganthan, Ponnuthurai Nagaratnam Wang, David Zhi Wei |
| author_sort | Du, Liang |
| collection | NTU |
| description | Among various time series (TS) forecasting methods, ensemble forecast is extensively acknowledged as a promising ensemble approach achieving great success in research and industry. Due to the high diversification of individual model assumptions, heterogeneous information fusion contributes to generating effective and robust forecasts for Economics, Meteorology, and Transportation. This paper proposes a Bayesian optimization-based dynamic ensemble (BODE) that overcomes the single model-based methods limitation and provides a dynamic ensemble forecast combination for TS with time-varying underlying patterns. The proposed BODE method combines ten disparate model candidates, including statistical methods, machine learning (ML)-based models, and the latest deep neural networks (DNN). We take into consideration their prediction performance for the recent past to adjust their weights for combination and apply the model-based Bayesian optimization algorithm (BOA) for the combination hyperparameter (HP) tuning to endow our method with higher adaptability and better generalization performance. Besides, the frequency impact of TS data on the ensemble forecast methods is under-researched in the current literature. Therefore, four groups of distinct seasonal TS datasets are investigated in this paper. The empirical result demonstrates that our method performs robustly better performance with the main reasons analyzed in a detailed ablation study. |
| first_indexed | 2024-10-01T04:29:29Z |
| format | Journal Article |
| id | ntu-10356/163881 |
| institution | Nanyang Technological University |
| language | English |
| last_indexed | 2024-10-01T04:29:29Z |
| publishDate | 2022 |
| record_format | dspace |
| spelling | ntu-10356/1638812022-12-21T02:47:57Z Bayesian optimization based dynamic ensemble for time series forecasting Du, Liang Gao, Ruobin Suganthan, Ponnuthurai Nagaratnam Wang, David Zhi Wei School of Civil and Environmental Engineering School of Electrical and Electronic Engineering Engineering::Civil engineering Time Series Forecasting Forecast Combination Among various time series (TS) forecasting methods, ensemble forecast is extensively acknowledged as a promising ensemble approach achieving great success in research and industry. Due to the high diversification of individual model assumptions, heterogeneous information fusion contributes to generating effective and robust forecasts for Economics, Meteorology, and Transportation. This paper proposes a Bayesian optimization-based dynamic ensemble (BODE) that overcomes the single model-based methods limitation and provides a dynamic ensemble forecast combination for TS with time-varying underlying patterns. The proposed BODE method combines ten disparate model candidates, including statistical methods, machine learning (ML)-based models, and the latest deep neural networks (DNN). We take into consideration their prediction performance for the recent past to adjust their weights for combination and apply the model-based Bayesian optimization algorithm (BOA) for the combination hyperparameter (HP) tuning to endow our method with higher adaptability and better generalization performance. Besides, the frequency impact of TS data on the ensemble forecast methods is under-researched in the current literature. Therefore, four groups of distinct seasonal TS datasets are investigated in this paper. The empirical result demonstrates that our method performs robustly better performance with the main reasons analyzed in a detailed ablation study. 2022-12-21T02:47:56Z 2022-12-21T02:47:56Z 2022 Journal Article Du, L., Gao, R., Suganthan, P. N. & Wang, D. Z. W. (2022). Bayesian optimization based dynamic ensemble for time series forecasting. Information Sciences, 591, 155-175. https://dx.doi.org/10.1016/j.ins.2022.01.010 0020-0255 https://hdl.handle.net/10356/163881 10.1016/j.ins.2022.01.010 2-s2.0-85123604243 591 155 175 en Information Sciences © 2022 Elsevier Inc. All rights reserved. |
| spellingShingle | Engineering::Civil engineering Time Series Forecasting Forecast Combination Du, Liang Gao, Ruobin Suganthan, Ponnuthurai Nagaratnam Wang, David Zhi Wei Bayesian optimization based dynamic ensemble for time series forecasting |
| title | Bayesian optimization based dynamic ensemble for time series forecasting |
| title_full | Bayesian optimization based dynamic ensemble for time series forecasting |
| title_fullStr | Bayesian optimization based dynamic ensemble for time series forecasting |
| title_full_unstemmed | Bayesian optimization based dynamic ensemble for time series forecasting |
| title_short | Bayesian optimization based dynamic ensemble for time series forecasting |
| title_sort | bayesian optimization based dynamic ensemble for time series forecasting |
| topic | Engineering::Civil engineering Time Series Forecasting Forecast Combination |
| url | https://hdl.handle.net/10356/163881 |
| work_keys_str_mv | AT duliang bayesianoptimizationbaseddynamicensemblefortimeseriesforecasting AT gaoruobin bayesianoptimizationbaseddynamicensemblefortimeseriesforecasting AT suganthanponnuthurainagaratnam bayesianoptimizationbaseddynamicensemblefortimeseriesforecasting AT wangdavidzhiwei bayesianoptimizationbaseddynamicensemblefortimeseriesforecasting |