A Hybrid Neural Network-Time Series Regression Model for Intermittent Demand Forecasting Data
DOI:
https://doi.org/10.52435/jaiit.v7i2.704Keywords:
Decision-making, exponential-smoothing, foreasting, Hybrid Model, Intermittent DataAbstract
Forecasting is a vital tool that helps us make informed decisions by predicting future events based on past data. For forecasts to be accurate, it is important that the data is reliable, complete, and consistent. Yet, the intermittent data is a unique data that is challenging to forecast. Intermittent data contains a characteristic that the data has a lot of long zeros in some periods. The zero value will influence the model to generate a forecasting model. This study aims to tackle those problems by applying a hybrid approach. We integrate the regression model and neural network to create a novel approach for forecasting intermittent data. The dataset used for this data is from Kaggle, sales at Walmart supermarket for one category only. The sales data always produce an intermittent demand pattern, because not every day are the items always sold to customers. This irregular pattern makes the data difficult to forecast using a naïve approach, such as the Croston method, exponential smoothing, and ARIMA. To evaluate the performance of our model, some metrics were calculated. We use mean squared error, root mean squared error, and root mean squared scaled error. The result shows that our proposed method outperforms the benchmark model, with an RMSSE of 0.98, which is the lowest compared to other benchmark models in the root mean squared scaled error value. This result shows promise as an exciting solution for overcoming the challenges posed by irregular data in future forecasting tasks.
References
S. Kolassa, "Commentary on the M5 forecasting competition," Int. J. Forecasting, vol. 38, no. 4, pp. 1562-1568, 2022, doi: 10.1016/j.ijforecast.2021.08.006
X. Tian, H. Wang, and E. E, “Forecasting intermittent demand for inventory management by retailers: A new approach,” J. Retail. Consum. Serv., 2021, doi: 10.1016/j.jretconser.2021.102662.
A. Muhaimin, E. Setyowati, K. M. H, A. Ruhui, and F. Sari, “Intermittent Data Forecasting using Kernel Support Vector Regression,” Int. Semin. Res. Mon. 2023. NST Proceedings., vol. 2024, pp. 23–26, 2024, doi: 10.11594/nstp.2024.4105.
R. Sarlo, C. Fernandes, and D. Borenstein, "Lumpy and intermittent retail demand forecasts with score-driven models," Eur. J. Oper. Res., vol. 307, no. 3, pp. 1146-1160, 2023, doi: 10.1016/j.ejor.2022.10.006. Available: https://www.sciencedirect.com/science/article/pii/S0377221722007743.
A. Muhaimin, P. Aji Riyantoko, H. Prabowo, and T. Trimono, “Negative Binomial Time Series Regression – Random Forest Ensemble in Intermittent Data,” Int. J. Data Sci. Eng. Anaylitics, 2021, doi: 10.33005/ijdasea.v1i2.10.
T. O. Hodson, "Root-mean-square error (RMSE) or mean absolute error (MAE): when to use them or not," Geosci. Model Dev., vol. 15, pp. 5481–5487, 2022, doi: 10.5194/gmd-15-5481-2022.
S. Kim and H. Kim, “A new metric of absolute percentage error for intermittent demand forecasts,” Int. J. Forecast., 2016, doi: 10.1016/j.ijforecast.2015.12.003.
J. M. Rozanec, B. Fortuna, and D. Mladenic, “Reframing Demand Forecasting: A Two-Fold Approach for Lumpy and Intermittent Demand,” Sustain., 2022, doi: 10.3390/su14159295.
A. Muhaimin, D. D. Prastyo, and H. H. S. Lu, “Forecasting with recurrent neural network in intermittent demand data,” in Proceedings of the Confluence 2021: 11th International Conference on Cloud Computing, Data Science and Engineering, 2021. doi: 10.1109/Confluence51648.2021.9376880.
P. Liu, “Intermittent demand forecasting for medical consumables with short life cycle using a dynamic neural network during the COVID-19 epidemic,” Health Informatics J., 2020, doi: 10.1177/1460458220954730.
P. G. Zhang, “Time series forecasting using a hybrid ARIMA and neural network model,” Neurocomputing, 2003, doi: 10.1016/S0925-2312(01)00702-0.
Suhartono, A. Choiruddin, H. Prabowo, and M. H. Lee, “Hybrid Machine Learning for Forecasting and Monitoring Air Pollution in Surabaya,” in Communications in Computer and Information Science, 2021. doi: 10.1007/978-981-16-7334-4_27.
N. A. B. Kamisan, M. H. Lee, S. Suhartono, A. G. Hussin, and Y. Z. Zubairi, “Load forecasting using combination model of multiple linear regression with neural network for Malaysian City,” Sains Malaysiana, 2018, doi: 10.17576/jsm-2018-4702-25.
S. Sharma, S. Sharma, and A. Athaiya, “Activation Functions In Neural Networks,” Int. J. Eng. Appl. Sci. Technol., 2020, doi: 10.33564/ijeast.2020.v04i12.054.
Suhartono, P. D. Saputri, F. F. Amalia, D. D. Prastyo, and B. S. S. Ulama, “Model selection in feedforward neural networks for forecasting inflow and outflow in Indonesia,” in Communications in Computer and Information Science, 2017. doi: 10.1007/978-981-10-7242-0_8.
S. Makridakis, E. Spiliotis, and V. Assimakopoulos, “M5 accuracy competition: Results, findings, and conclusions,” Int. J. Forecast., 2022, doi: 10.1016/j.ijforecast.2021.11.013.
R. Shah, P. Shah, C. Joshi, R. Jain and R. Nikam, "Linear Regression vs LSTM for Time Series Data," 2022 IEEE World Conference on Applied Intelligence and Computing (AIC), Sonbhadra, India, 2022, pp. 670-675, doi: 10.1109/AIC55036.2022.9848887.
É. Perthame, F. Forbes, B. Olivier, and A. Deleforge, "Non linear robust regression in high dimension," in Proc. [Conference Name], 2016. Available: https://api.semanticscholar.org/CorpusID:36302546.
S. Sharma, S. Sharma, and A. Athaiya, "Activation functions in neural networks," Int. J. Eng. Appl. Sci. Technol., vol. 4, pp. 310-316, May 2020, doi: 10.33564/IJEAST.2020.v04i12.054.
Suhartono, M. H. Lee, and D. D. Prastyo, “Two levels ARIMAX and regression models for forecasting time series data with calendar variation effects,” in AIP Conference Proceedings, 2015. doi: 10.1063/1.4937108.
R. J. Hyndman and A. B. Koehler, “Another look at measures of forecast accuracy,” Int. J. Forecast., 2006, doi: 10.1016/j.ijforecast.2006.03.001.
W. Wei, Time Series Analysis: Univariate and Multivariate Methods Second Edition. United States of America: Pearson Education, Inc., 2006.
H. Narayanan, T. Seidler, M. F. Luna, M. Sokolov, M. Morbidelli, and A. Butté, "Hybrid models for the simulation and prediction of chromatographic processes for protein capture," J. Chromatogr. A, vol. 1650, p. 462248, 2021, doi: 10.1016/j.chroma.2021.462248.
N. G. Reich, J. Lessler, K. Sakrejda, S. A. Lauer, S. Iamsirithaworn, and D. A. T. Cummings, "Case study in evaluating time series prediction models using the relative mean absolute error," The Amer. Stat., vol. 70, no. 3, pp. 285–292, 2016, doi: 10.1080/00031305.2016.1148631.
Downloads
Published
How to Cite
Issue
Section
License
Copyright (c) 2025 Amri Muhaimin, Aviolla Terza Damaliana, Muhammad Nasrudin, Prismahardi Aji Riyantoko, Nabilah Selayanti, Shafira Amanda Putri
This work is licensed under a Creative Commons Attribution-ShareAlike 4.0 International License.
All our articles are published under a Creative Commons Attribution-ShareAlike 4.0 International (CC BY-SA 4.0) license.
