Trend-based Dynamic Selection of Machine Learning Models for Pandemic Time Series Forecasting

• Eduardo H. X. de M. e Menezes UFPE
• Jair P. de Sales UFCA
• Paulo S. G. de Mattos Neto UFPE

DOI: https://doi.org/10.5753/eniac.2025.13836

Abstract

Pandemic time series forecasting is challenging due to rapidly changing patterns, which general-purpose predictors often fail to capture. Recent studies suggest that specialized models are better suited for such complexity, especially when trend changes are key. This work proposes a forecasting approach that dynamically selects specialized Support Vector Regression (SVR) models based on trend classification (SVRTC). SVRTC maintains a pool of models, each tailored to exponential growth, plateau, or decline patterns. Using COVID-19 case series from eight countries, SVRTC was compared to traditional forecasting and existing dynamic selection methods. It achieved remarkable performance among traditional Machine Learning models and competitive results against state-of-the-art dynamic selection, with low computational cost, indicating strong potential for broader application.

References

Ahmadi, M. and Khashei, M. (2021). Generalized Support Vector Machines (GSVMs) Model for Real-World Time Series Forecasting. Soft Computing, 25(22):14139–14154.

Ahmetolan, S., Bilge, A. H., Demirci, A., Peker-Dobie, A., and Ergonul, O. (2020). What Can We Estimate From Fatality and Infectious Case Data Using the Susceptible-Infected-Removed (SIR) Model? A Case Study of Covid-19 Pandemic. Frontiers in Medicine, 7.

Botz, J., Valderrama, D., Guski, J., and Fröhlich, H. (2024). A Dynamic Ensemble Model for Short-Term Forecasting in Pandemic Situations. PLOS Global Public Health, 4(8):e0003058.

Chung, H. W., Apio, C., Goo, T., Heo, G., Han, K., Kim, T., Kim, H., Ko, Y., Lee, D., Lim, J., et al. (2021). Effects of Government Policies on the Spread of COVID-19 Worldwide. Scientific Reports, 11(1):20495.

de Oliveira, J. F. L., Silva, E. G., and de Mattos Neto, P. S. G. (2022). A Hybrid System Based on Dynamic Selection for Time Series Forecasting. IEEE Transactions on Neural Networks and Learning Systems, 33(8):3251–3263.

de Sales, J. P., de Carvalho Lima, J. E., Firmino, P. R. A., and de Mattos Neto, P. S. (2023). Oceanic Niño Index Forecasting Based on Dynamic Ensemble Selection. In Simpósio Brasileiro de Pesquisa Operacional, São José dos Campos -SP.

de Sales, J. P., de Mattos Neto, P. S., and Firmino, P. R. (2024). A Dynamic Ensemble Approach Based on Trend Analysis to COVID-19 Incidence Forecast. Biomedical Signal Processing and Control, 95:106435.

Firmino, P. R. A., De Sales, J. P., Júnior, J. G., and Da Silva, T. A. (2020). A Non-Central Beta Model to Forecast and Evaluate Pandemics Time Series. Chaos, Solitons & Fractals, 140.

He, H., Zhang, Q., Yi, K., Shi, K., Niu, Z., and Cao, L. (2025). Distributional Drift Adaptation With Temporal Conditional Variational Autoencoder for Multivariate Time Series Forecasting. IEEE Transactions on Neural Networks and Learning Systems, 36(4):7287–7301.

Hu, J. and Zheng, W. (2020). A Deep Learning Model to Effectively Capture Mutation Information in Multivariate Time Series Prediction. Knowledge-Based Systems, 203:106139.

Hyndman, R. J. and Athanasopoulos, G. (2018). Forecasting: Principles and Practice. OTexts.

Ioannidis, J. P., Cripps, S., and Tanner, M. A. (2022). Forecasting For COVID-19 Has Failed. International Journal of Forecasting, 38(2):423–438.

Kaur, J., Parmar, K. S., and Singh, S. (2023). Autoregressive Models in Environmental Forecasting Time Series: A Theoretical and Application Review. Environmental Science and Pollution Research, 30(8):19617–19641.

Li, Y., Zhu, Z., Kong, D., Han, H., and Zhao, Y. (2019). EA-LSTM: Evolutionary Attention-Based LSTM for Time Series Prediction. Knowledge-Based Systems, 181:104785.

Liapis, C. M., Karanikola, A., and Kotsiantis, S. (2020). An Ensemble Forecasting Method Using Univariate Time Series COVID-19 Data. In Proceedings of the 24th Pan-Hellenic Conference on Informatics, pages 50–52.

Maaliw, R. R., Ballera, M. A., Mabunga, Z. P., Mahusay, A. T., Dejelo, D. A., and Seño, M. P. (2021). An Ensemble Machine Learning Approach for Time Series Forecasting of COVID-19 Cases. In 2021 IEEE 12th Annual Information Technology, Electronics and Mobile Communication Conference (IEMCON), pages 0633–0640. IEEE.

Machado, J. T. and Lopes, A. M. (2020). Rare and Extreme Events: The Case of COVID-19 Pandemic. Nonlinear Dyn, 100:2953–2972.

Qu, Z., Li, Y., Jiang, X., and Niu, C. (2023). An Innovative Ensemble Model Based on Multiple Neural Networks and a Novel Heuristic Optimization algorithm for COVID-19 forecasting. Expert Systems with Applications, 212:118746.

Rahimi, I., Chen, F., and Gandomi, A. H. (2023). A Review on COVID-19 Forecasting Models. Neural computing and applications, 35(33):23671–23681.

Saadallah, A. (2023). Online Explainable Model Selection for Time Series Forecasting. In 2023 IEEE 10th International Conference on Data Science and Advanced Analytics (DSAA), pages 1–10.

Saadallah, A., Jakobs, M., and Morik, K. (2022). Explainable Online Ensemble of Deep Neural Network Pruning for Time Series Forecasting. Machine Learning, 111:3459–3487.

Santos, W. R., Sampaio Jr, A. R., Rosa, N. S., and Cavalcanti, G. D. (2024). Microservices Performance Forecast Using Dynamic Multiple Predictor Systems. Engineering Applications of Artificial Intelligence, 129:107649.

Sheikh, M. R. and Coulibaly, P. (2025). Introducing time series features based dynamic weights estimation framework for hydrologic forecast merging. Journal of Hydrology, 654:132872.

Silva, E. G., Cavalcanti, G. D. C., de Oliveira, J. F. L., and de Mattos Neto, P. S. G. (2020). On the Evaluation of Dynamic Selection Parameters for Time Series Forecasting. In 2020 International Joint Conference on Neural Networks (IJCNN), pages 1–7.

Silva, E. G., De Mattos Neto, P. S. G., and Cavalcanti, G. D. C. (2021). A Dynamic Predictor Selection Method Based on Recent Temporal Windows for Time Series Forecasting. IEEE Access, 9:108466–108479.

Stasiak, M. D. and Staszak, Ż. (2024). Modelling and forecasting crude oil prices using trend analysis in a binary-temporal representation. Energies, 17(14):3361.

Tang, Y., Yu, F., Pedrycz, W., Yang, X., Wang, J., and Liu, S. (2021). Building trend fuzzy granulation-based lstm recurrent neural network for long-term time-series forecasting. IEEE transactions on fuzzy systems, 30(6):1599–1613.

Tealab, A. (2018). Time Series Forecasting Using Artificial Neural Networks Methodologies: A Systematic Review. Future Computing and Informatics Journal, 3(2):334–340.

Tratar, L. F., Mojškerc, B., and Toman, A. (2016). Demand Forecasting with Four-Parameter Exponential Smoothing. International Journal of Production Economics, 181:162–173.

Yao, C., Dai, Q., and Song, G. (2019). Several Novel Dynamic Ensemble Selection Algorithms for Time Series Prediction. Neural Processing Letters, 50:1789–1829.

Zeng, A., Chen, M., Zhang, L., and Xu, Q. (2023). Are Transformers Effective for Time Series Forecasting? In Proceedings of the AAAI conference on artificial intelligence, volume 37, pages 11121–11128.