Comparative Analysis of Time Series Forecasting Methods for Predicting Spare Parts Demand in Supply Chains

• Ítalo P. Caliari Fundação Paulo Feitoza
• Ravi B. D. Figueiredo Fundação Paulo Feitoza
• Diogenes W. F Silva Fundação Paulo Feitoza
• Edierley B. Messias Fundação Paulo Feitoza
• Mario Haddad Neto Fundação Paulo Feitoza

DOI: https://doi.org/10.5753/semish.2026.21311

Resumo

Accurate demand forecasting prevents stockouts and excess inventory, but sparse data for many components hampers model building. To address this, interchangeable components were grouped, enabling analysis of aggregated warranty, repair, and Consumption Index (CI) data for 3,000 groups. Classical time series techniques, including the Simple Moving Average (SMA) and Exponential Moving Average (EMA) were evaluated along with different SARIMAX implementations. Performance analysis was conducted using the Root Mean Square Error (RMSE) and Sufficiency metrics. The results of this study highlight the effectiveness of simple approaches compared to more complex methods, such as neural networks and XGBoost, which did not outperform baseline algorithms in terms of RMSE and Sufficiency. However, simpler methods have limitations in capturing complex dynamics.

Referências

Ameer, T. and Fatahi Valilai, O. (2025). Predictive exploratory data analysis of shopfloor cnc machine operation through a machine learning model. Journal of Open Innovation: Technology, Market, and Complexity, 11(2):100559.

Andrianakis, I., Gkatas, V., Eleftheriadis, N., Ellinidis, A., and Avramidou, E. (2024). Predictionscms: The implementation of an ai-powered supply chain management system. International Journal of Industrial and Manufacturing Engineering, 18(3):65 – 71.

Box, G. E. P., Jenkins, G. M., Reinsel, G. C., and Ljung, G. M. (2015). Time Series Analysis: Forecasting and Control. 5 edition.

Fanoodi, B., Malmir, B., and Jahantigh, F. F. (2019). Reducing demand uncertainty in the platelet supply chain through artificial neural networks and arima models. Computers in Biology and Medicine, 113:103415.

Hansun, S. (2013). A new approach of moving average method in time series analysis. In 2013 Conference on New Media Studies (CoNMedia), pages 1–4.

Hyndman, R. and Athanasopoulos, G. (2021). Forecasting: principles and practice,. OTexts: Melbourne, Australia.

Ji, W. and Wang, L. (2017). Big data analytics based fault prediction for shop floor scheduling. Journal of Manufacturing Systems, 43:187–194.

Kherdekar, V. A. and Naik, S. (2024). Impact of feature normalization techniques for recognition of speech for mathematical expression. In Senjyu, T., So-In, C., and Joshi, A., editors, Smart Trends in Computing and Communications, pages 109–117, Singapore. Springer Nature Singapore.

Klinker, F. (2011). Exponential moving average versus moving exponential average. Math. Semesterber., 58(1):97–107.

Liemohn, M. W., Shane, A. D., Azari, A. R., Petersen, A. K., Swiger, B. M., and Mukhopadhyay, A. (2021). Rmse is not enough: Guidelines to robust data-model comparisons for magnetospheric physics. Journal of Atmospheric and Solar-Terrestrial Physics, 218:105624.

Rahman Mahin, M. P., Shahriar, M., Das, R. R., Roy, A., and Reza, A. W. (2025). Enhancing sustainable supply chain forecasting using machine learning for sales prediction. Procedia Computer Science, 252:470–479. 4th International Conference on Evolutionary Computing and Mobile Sustainable Networks.

Sierra Espinel, A. I. and Suarez Barón, M. J. (2025). Applying deep learning and forecasting techniques to the pharmaceutical supply chain. Procedia Computer Science, 253:2791–2800. 6th International Conference on Industry 4.0 and Smart Manufacturing.

SJ, T. and B., L. (2017). Forecasting at scale. PeerJ Preprints, 5:e3190v2.

Swamidass, P. M. (2000). INTERCHANGEABILITY, pages 293–294. Springer US, New York, NY.