Development of machine learning based demand forecasting models for the e-commerce sector

Öz

The e-commerce sector has undergone rapid and dynamic growth in recent years. For companies aspiring to lead in this competitive industry, it is crucial to efficiently and cost-effectively respond to evolving consumer demands. In this context, the ability to accurately forecast future product demand becomes imperative. This study aims to develop forecasting models utilizing machine learning-based techniques, specifically Multi-Layer Perceptron (MLP), Multi-Horizon Quantile Recurrent Neural Network (MQRNN), and Random Forest (RF), to predict future product demand. The demand forecasting models were developed for the months of July and August, based on daily sales data for Fast-Moving Consumer Goods (FMCG) products spanning from January 1, 2023, to August 25, 2024. The models’ performances were evaluated using Mean Absolute Percentage Error (MAPE). Upon examining the forecasting models developed using MLP, MQRNN, and RF, it has been observed that MQRNN exhibited the superior performance.

Anahtar Kelimeler

• E-Commerce
• Demand Forecasting
• Machine Learning

E-ticaret sektörü için makine öğrenimi tabanlı talep tahmin modellerinin geliştirilmesi

Öz

E-ticaret sektörü son yıllarda hızlı ve dinamik bir büyüme göstermiştir. Bu rekabetçi sektörde lider olmayı hedefleyen şirketler için değişen tüketici taleplerine verimli ve maliyet etkin bir şekilde yanıt vermek büyük önem taşımaktadır. Bu bağlamda, gelecekteki ürün talebini doğru bir şekilde tahmin etme yeteneği hayati hale gelmektedir. Bu çalışma, gelecekteki ürün talebini tahmin etmek amacıyla, Çok Katmanlı Algılayıcı (MLP), Çok Ufuklu Çeyrek Tekrarlayan Sinir Ağı (MQRNN) ve Rastgele Orman (RF) gibi makine öğrenimi tabanlı teknikler kullanılarak tahmin modelleri geliştirmeyi amaçlamaktadır. Hızlı Tüketim Ürünleri (FMCG) için günlük satış verilerine dayalı olarak 1 Ocak 2023 ile 25 Ağustos 2024 tarihleri arasındaki dönemi kapsayan bu modeller, Temmuz ve Ağustos aylarına yönelik talep tahmini yapmak için oluşturulmuştur. Modellerin performansları Ortalama Mutlak Yüzde Hatası (MAPE) metriği kullanılarak değerlendirilmiştir. MLP, MQRNN ve RF kullanılarak geliştirilen tahmin modelleri incelendiğinde, en iyi performansı MQRNN modelinin gösterdiği gözlemlenmiştir.

Anahtar Kelimeler

• E-Ticaret
• Talep Tahmini
• Makine Öğrenimi

Kaynakça

1. Ahmadov, Y., & Helo, P. (2023). Deep learning-based approach for forecasting intermittent online sales. Discover Artificial Intelligence, 3(1), 45.
2. Aravazhi, A. (2021). Hybrid machine learning models for forecasting surgical case volumes at a hospital. AI, 2(4), 512-526.
3. Chen, Y., Xie, X., Pei, Z., Yi, W., Wang, C., Zhang, W., & Ji, Z. (2024). Development of a Time Series E-Commerce Sales Prediction Method for Short-Shelf-Life Products Using GRU-LightGBM. Applied Sciences, 14(2), 866.
4. Chi, Y., Lei, D., Zheng, L., & Shen, Z. J. M. (2024). Demand Forecasting During Grand Promotion for Online Retailing. Available at SSRN 4777632.
5. Daulat Desale, I. (2024). E-commerce Sales Forecasting Using Machine Learning Algorithm (Doctoral dissertation, Dublin Business School).
6. Ecevit, A., Öztürk, İ., Dağ, M., & Özcan, T. (2024). Short-Term Sales Forecasting Using LSTM and Prophet Based Models in E-Commerce. Acta Infologica, 7(1), 59-70.
7. Febima, M., & Magdalena, L. (2024). Predictive Analytics on Shopee for Optimizing Product Demand Prediction through K-Means Clustering and KNN Algorithm Fusion. Journal of Information Systems and Informatics, 6(2), 751-765.
8. Islam, M. T., Ayon, E. H., Ghosh, B. P., MD, S. C., Shahid, R., Rahman, S., ... & Nguyen, T. N. (2024). Revolutionizing Retail: A Hybrid Machine Learning Approach for Precision Demand Forecasting and Strategic Decision-Making in Global Commerce. Journal of Computer Science and Technology Studies, 6(1), 33-39.

9. Ivanov, D. (2024). Demand Forecasting, Production Planning, and Inventory Control. In Introduction to Supply Chain Analytics: With Examples in AnyLogic and anyLogistix Software (pp. 21-47). Cham: Springer Nature Switzerland.
10. Kaunchi, P., Jadhav, T., Dandawate, Y., & Marathe, P. (2021, October). Future Sales Prediction For Indian Products Using Convolutional Neural Network-Long Short Term Memory. In 2021 2nd Global Conference for Advancement in Technology (GCAT) (pp. 1-5). IEEE.
11. Lu, T. (2024). Research on sales forecasting in e-commerce industry for imbalanced classification data. Science and Technology of Engineering, Chemistry and Environmental Protection, 1(6).
12. Mejía, S., & Aguilar, J. (2024). A demand forecasting system of product categories defined by their time series using a hybrid approach of ensemble learning with feature engineering. Computing, 1-21.
13. Nasseri, M., Falatouri, T., Brandtner, P., & Darbanian, F. (2023). Applying Machine Learning in Retail Demand Prediction—A Comparison of Tree-Based Ensembles and Long Short-Term Memory-Based Deep Learning. Applied Sciences, 13(19), 11112.
14. Neelakandan, S., Prakash, V., PranavKumar, M. S., & Balasubramaniam, R. (2023, June). Forecasting of E-Commerce System for Sale Prediction Using Deep Learning Modified Neural Networks. In 2023 International Conference on Applied Intelligence and Sustainable Computing (ICAISC) (pp. 1-5). IEEE.
15. Park, S., & Kim, J., Landslide susceptibility mapping based on random forest and boosted regression tree models, and a comparison of their performance. Applied Sciences, 9(5), 942, 2019.
16. Pavlyshenko, B. M. (2022). Forecasting of Non-Stationary Sales Time Series Using Deep Learning. arXiv preprint arXiv:2205.11636.
17. Rosário, A., & Raimundo, R. (2021). Consumer marketing strategy and E-commerce in the last decade: a literature review. Journal of theoretical and applied electronic commerce research, 16(7), 3003-3024.
18. Singh, S., & Vijay, T. S. (2024). Technology roadmapping for the e-commerce sector: A text-mining approach. Journal of Retailing and Consumer Services, 81, 103977.
19. Swaminathan, K., & Venkitasubramony, R. (2024). Demand forecasting for fashion products: A systematic review. International Journal of Forecasting, 40(1), 247-267.
20. Wang, J., Chong, W. K., Lin, J., & Hedenstierna, C. P. T. (2023). Retail Demand Forecasting Using Spatial-Temporal Gradient Boosting Methods. Journal of Computer Information Systems, 1-13.
21. Wu, P., Zhang, G., Li, Y., & Chen, X. (2023). Research on E-Commerce Inventory Demand Forecasting Based on NAR Neural Network. Open Access Library Journal, 10(5), 1-11.
22. Zhang, X. Y., Watkins, C., & Kuenzel, S., Multi-quantile recurrent neural network for feeder-level probabilistic energy disaggregation considering roof-top solar energy. Engineering Applications of Artificial Intelligence, 110, 104707, 2022.