Bir Çimento Firmasında İstatistiksel Zaman Serileri Yöntemleri ve Derin Öğrenme ile Talep Tahminleme

Abstract

Talep tahminleri, üretim planlama, finansal planlama, bütçeleme, satın alma, satış gibi birçok iş sürecinin yönetilebilmesi için kritik öneme sahiptir. Talep tahminlerinin yüksek doğrulukla elde edilmesi, tüm tedarik zinciri yönetimi süreçlerinin başarısı için kilit bir faktördür. Bu çalışmada talep tahminleme problemi, bir çimento firmasının en yüksek satış payına sahip bir ürün grubunun geçmiş satış verileri kullanılarak ele alınmıştır. Ele alınan veri seti tek değişkenli bir zaman serisidir. Veri setinin ilk üç yılı eğitim, son yılı ise test seti olarak kullanılmıştır. Tahminleme için öncelikle geleneksel istatistiksel zaman serileri analiz yöntemleri uygulanmıştır. Eğitim setinde, uygulanan yöntemler içinde en başarılı olan istatistiksel zaman serileri yöntemi Basit Mevsimsel Yöntem (Simple Seasonal Method - SSM) olmuştur. SSM modelinin performansı, derin öğrenme yöntemlerinden Uzun-Kısa Süreli Bellek (Long-Short Term Memory-LSTM) temelli olarak geliştirilen modelin performansıyla karşılaştırılmıştır. LSTM modeli geliştirilirken ızgara (grid) arama yapılmış ve hiper-parametrelerin değerleri için en başarılı kombinasyon belirlenmiştir. Bu konfigürasyonla eğitilen LSTM modeli test setinde uygulanmıştır. Modellerin test setindeki tahmin performansları karşılaştırıldığında, LSTM modelinin SSM modeline göre MAPE ölçütünde %34,57, RMSE ölçütünde ise %33,74 iyileştirme sağladığı görülmüştür.

Keywords

• Talep tahminleme
• Tek değişkenli zaman serileri
• İstatistiksel zaman serileri yöntemleri
• Derin öğrenme
• , Derin öğrenme

Demand Forecasting with Statistical Time Series Analysis Methods and Deep Learning in a Cement Firm

Abstract

Demand forecasts are critical for managing various business processes, including production planning, financial planning, budgeting, purchasing, and sales. Therefore, obtaining demand forecasts with high accuracy is a critical factor in successfully managing all supply chain management processes. This study addresses the demand forecasting problem by using actual past sales data of a cement firm’s one product group, which has the largest sales share among all product groups. The handled data set is a univariate time series. The first three years of the data set are used as the training set; the remaining one-year data is used as the test set. Firstly, classical statistical time series analysis methods are applied to the training data. Among the methods, the Simple Seasonal Method (SSM) has the best performance. The performance of the SSM model is compared to the model developed based on Long-Short Term Memory (LSTM), a deep learning method well-known for its success for time series data. A grid search approach is performed to determine the best combination of the values of hyper-parameters for LSTM models. Finally, the selected LSTM model configuration is applied to the test set. The prediction performances of the models in the test set indicate that the LSTM model provides 34.57% improvement in the MAPE criterion and 33.74% in the RMSE criterion compared to the SSM model.

Keywords

• Demand forecasting
• Univariate time series
• Statistical time series analysis methods
• Deep learning
• LSTM

References

1. APICS Dictionary (1987), American Production and Inventory Control Society, Inc., Falls Church, VA.
2. Merkuryeva, G., Valberga, A., & Smirnov, A. (2019). Demand forecasting in pharmaceutical supply chains: A case study. Procedia Computer Science, 149, 3-10.
3. Stevenson, W. J. (2012). Operations Management. 11th Edition. McGraw-Hill Global Education.
4. Lawrence, K. D., & Klimberg, R.K. (Eds.). (2018). Advances in business and management forecasting. Volume 12. Emerald Group Publishing.
5. Box, G. E., Jenkins, G. M., Reinsel, G. C., & Ljung, G. M. (2015). Time series analysis: forecasting and control. John Wiley & Sons.
6. Bontempi, G., Ben Taieb, S., & Borgne, Y. A. L. (2012, July). Machine learning strategies for time series forecasting. In European business intelligence summer school (pp. 62-77). Springer, Berlin, Heidelberg.
7. Loureiro, A. L., Miguéis, V. L., & da Silva, L. F. (2018). Exploring the use of deep neural networks for sales forecasting in fashion retail. Decision Support Systems, 114, 81-93.
8. Wanchoo, K. (2019, March). Retail demand forecasting: a comparison between deep neural network and gradient boosting method for univariate time series. In 2019 IEEE 5th International Conference for Convergence in Technology (I2CT) (pp. 1-5). IEEE.

9. Mahmoud, A., & Mohammed, A. (2021). A survey on deep learning for time-series forecasting. In Machine Learning and Big Data Analytics Paradigms: Analysis, Applications and Challenges (pp. 365-392). Springer, Cham.
10. Ramos, P., Santos, N., & Rebelo, R. (2015). Performance of state space and ARIMA models for consumer retail sales forecasting. Robotics and computer-integrated manufacturing, 34, 151-163.
11. Pongdatu, G. A. N., & Putra, Y. H. (2018, August). Seasonal time series forecasting using SARIMA and Holt Winter’s exponential smoothing. In IOP Conference Series: Materials Science and Engineering (Vol. 407, No. 1, p. 012153). IOP Publishing.
12. Bandara, K., Shi, P., Bergmeir, C., Hewamalage, H., Tran, Q., & Seaman, B. (2019, December). Sales demand forecast in e-commerce using a long short-term memory neural network methodology. In International conference on neural information processing (pp. 462-474). Springer, Cham.
13. Abbasimehr, H., Shabani, M., & Yousefi, M. (2020). An optimized model using LSTM network for demand forecasting. Computers & industrial engineering, 143, 106435.
14. Pacella, M., & Papadia, G. (2021). Evaluation of deep learning with long short-term memory networks for time series forecasting in supply chain management. Procedia CIRP, 99, 604-609.
15. Ensafi, Y., Amin, S. H., Zhang, G., & Shah, B. (2022). Time-series forecasting of seasonal items sales using machine learning–A comparative analysis. International Journal of Information Management Data Insights, 2(1), 100058.
16. Goh, C., & Law, R. (2002). Modeling and forecasting tourism demand for arrivals with stochastic nonstationary seasonality and intervention. Tourism management, 23(5), 499-510.
17. Chase, C. W. (2013). Demand-driven forecasting: a structured approach to forecasting. John Wiley & Sons.
18. Dankwa, P., Cudjoe, E., Amuah, E. E. Y., Kazapoe, R. W., & Agyemang, E. P. (2021). Analyzing and forecasting rainfall patterns in the Manga-Bawku area, northeastern Ghana: Possible implication of climate change. Environmental Challenges, 5, 100354.
19. Chatfield, C., & Yar, M. (1988). Holt‐Winters forecasting: some practical issues. Journal of the Royal Statistical Society: Series D (The Statistician), 37(2), 129-140.
20. Zhang, G. P. (2003). Time series forecasting using a hybrid ARIMA and neural network model. Neurocomputing, 50, 159-175.
21. Hochreiter, S., & Schmidhuber, J. (1997). Long short-term memory. Neural computation, 9(8), 1735-1780.
22. Bandara, K., Bergmeir, C., & Smyl, S. (2020). Forecasting across time series databases using recurrent neural networks on groups of similar series: A clustering approach. Expert systems with applications, 140, 112896.
23. Khalil, K., Eldash, O., Kumar, A., & Bayoumi, M. (2019). Economic LSTM approach for recurrent neural networks. IEEE Transactions on Circuits and Systems II: Express Briefs, 66(11), 1885-1889.