Call Arrival Forecasting Using Feature Selection and Bayesian Optimization in Machine Learning and Deep Learning Models

Abstract

With the increasing complexity and dynamic nature of call center operations, accurate call arrival prediction has emerged as a significant research area, attracting attention from both academia and industry. Forecasting call arrivals is essential for effective resource allocation, staffing decisions, and service level planning, ultimately contributing to improved operational efficiency and customer satisfaction. study investigates the effectiveness of combined feature selection and Bayesian optimization techniques in enhancing call arrival prediction accuracy. We conducted a comprehensive analysis of several ML and DL models, our experiments compared model performance using a combined feature selection and model optimization strategy. Utilizing real-world data from call center operations, we employed three datasets with daily, hourly, and half-hourly observations to predict call volume and Average Handling Time (AHT). Our findings revealed a significant enhancement in model efficiency and predictive accuracy with the integration of Bayesian optimization. Specifically, for call volume predictions, the most dramatic decrease in Mean Absolute Error (MAE) was observed with hourly intervals, where the CNN model with selected features achieved a reduction from 0.74 to 0.02, marking a 97.3% error reduction. For Average Handle Time (AHT), the most notable improvement was seen with half-hourly aggregations, where the CNN model with selected features showed a reduction from 0.78 to 0.02, resulting in a 97.44% error reduction. These results highlight the effectiveness of combining feature selection with Bayesian optimization methods for more accurate predictions of both call volume and AHT. Feature selection not only improved the performance of BO-optimized models but also provided valuable insights into the most relevant features for different predictive tasks. Furthermore, our findings emphasize the importance of using shorter time intervals, such as hourly and half-hourly aggregations, for improved prediction accuracy. This dual approach emphasizes the potential of advanced optimization and feature selection techniques in enhancing predictive modeling accuracy. These findings are important for businesses that depend on efficient call center operations, as they demonstrate the advantages of using advanced optimization techniques and sophisticated modeling methods

Keywords

• Forecasting
• Call Arrivals
• Bayesian Optimization
• Feature Selection

References

1. [1] Gans, N., Koole, G., Mandelbaum, A., (2003). Telephone Call Centers: Tutorial, Review, and Research Prospects. Manufacturing and Service Operations Management. 5(2): 79–141. doi: 10.1287/msom.5.2.79.16071.
2. [2] Bhulai, S., Koole, G., Pot, A., (2008). Simple methods for shift scheduling in multiskill call centers. Manufacturing and Service Operations Management. 10(3): 411–20. doi: 10.1287/msom.1070.0172.
3. [3] Bastianin, A., Galeotti, M., Manera, M., (2012). Forecast Evaluation in Call Centers: Combined Forecasts, Flexible Loss Functions and Economic Criteria. SSRN Electronic Journal.: 1–33. doi: 10.2139/ssrn.1829891.
4. [4] Martin, R.J., Mousavi, R., Saydam, C., (2021). Predicting emergency medical service call demand: A modern spatiotemporal machine learning approach. Operations Research for Health Care. 28: 100285.
5. [5] Kanthanathan, C., Carty, G., Raja, M.A., Ryan, C., (2020). Recurrent Neural Network based Automated Workload Forecasting in a Contact Center. 2020 3rd International Conference on Intelligent Sustainable Systems (ICISS), p. 1423–8.
6. [6] Manno, A., Rossi, F., Smriglio, S., Cerone, L., (2023). Comparing deep and shallow neural networks in forecasting call center arrivals. Soft Computing. 27(18): 12943–57. doi: 10.1007/s00500-022-07055-2.
7. [7] Chacón, H., Koppisetti, V., Hardage, D., Choo, K.K.R., Rad, P., (2023). Forecasting call center arrivals using temporal memory networks and gradient boosting algorithm. Expert Systems with Applications. 224(April): 119983. doi: 10.1016/j.eswa.2023.119983.
8. [8] Ibrahim, R., Ye, H., L’Ecuyer, P., Shen, H., (2016). Modeling and forecasting call center arrivals: A literature survey and a case study. International Journal of Forecasting. 32(3): 865–74.

9. [9] Baskaran, T., John, N., Dhandra, B. V., (2023). Hybrid Model Using Interacted-ARIMA and ANN Models for Efficient Forecasting BT - Multi-disciplinary Trends in Artificial Intelligence. In: Morusupalli, R., Dandibhotla, T.S., Atluri, V.V., Windridge, D., Lingras, P., Komati, V.R., editors. Cham: Springer Nature Switzerland p. 747–56.
10. [10] Soy Temür, A., Yıldız, Ş., (2021). Comparison of Forecasting Performance of ARIMA LSTM and HYBRID Models for The Sales Volume Budget of a Manufacturing Enterprise. Istanbul Business Research. 50(1): 15–46.
11. [11] Barrow, D.K., (2016). Forecasting intraday call arrivals using the seasonal moving average method. Journal of Business Research. 69(12): 6088–96. doi: https://doi.org/10.1016/j.jbusres.2016.06.016.
12. [12] Ballouch, M., Akay, F., Erdem, S., Tartuk, M., Nurdağ, T.F., Yurdagül, H.H., (2021). Forecasting Call Center Arrivals Using Machine Learning. Osmaniye Korkut Ata Üniversitesi Fen Bilimleri Enstitüsü Dergisi. 4(1): 96–101. doi: 10.47495/okufbed.824870.
13. [13] Albrecht, T., Rausch, T.M., Derra, N.D., (2021). Call me maybe: Methods and practical implementation of artificial intelligence in call center arrivals’ forecasting. Journal of Business Research. 123: 267–78.
14. [14] Kadioglu, M.A., Alatas, B., (2023). Enhancing Call Center Efficiency: Data Driven Workload Prediction and Workforce Optimization. Eurasia Proceedings of Science, Technology, Engineering and Mathematics. 24: 96–100. doi: 10.55549/epstem.1406245.
15. [15] Andrade, R., Moazeni, S., (2023). Transfer rate prediction at self-service customer support platforms in insurance contact centers. Expert Systems with Applications. 212: 118701. doi: https://doi.org/10.1016/j.eswa.2022.118701.
16. [16] Torres, J.F., Hadjout, D., Sebaa, A., Martínez-Álvarez, F., Troncoso, A., (2021). Deep Learning for Time Series Forecasting: A Survey. Big Data. 9(1): 3–21. doi: 10.1089/big.2020.0159.
17. [17] Abut, F., Tartuk, M., Nurdağ, T.F., Acar, V., Erdem, S., Akay, F., (2022). Forecasting Call Center Arrivals Using XGBoost Combined with Consecutive and Periodic Lookback. Eastern Anatolian Journal of Science. 8(1): 20–5.
18. [18] Kiwamu, Y., Goro, H., (2019). Forecasting call arrivals at call center using dynamic linear model 51: 1–7.
19. [19] Kumwilaisak, W., Phikulngoen, S., Piriyataravet, J., Thatphithakkul, N., Hansakunbuntheung, C., (2022). Adaptive Call Center Workforce Management With Deep Neural Network and Reinforcement Learning. IEEE Access. 10: 35712–24. doi: 10.1109/ACCESS.2022.3160452.
20. [20] Surasai, P., Sa-ing, V., (2023). Time Series Forecast Of Call Arrivals Using Machine Learning Methods: 273–92.
21. [21] Shcherbakov, M.V., Brebels, A., Shcherbakova, N.L., Tyukov, A.P., Janovsky, T.A., Kamaev, V.A., (2013). A Survey of Forecast Error Measures. World Applied Sciences Journal. 24((Information Technologies in Modern Industry, Education & Society)): 171–6. doi: 10.5829/idosi.wasj.2013.24.itmies.80032.
22. [22] Li, X., Chen, W., Zhang, Q., Wu, L., (2020). Building Auto-Encoder Intrusion Detection System based on random forest feature selection. Computers & Security. 95: 101851. doi: https://doi.org/10.1016/j.cose.2020.101851.
23. [23] Niu, D., Wang, K., Sun, L., Wu, J., Xu, X., (2020). Short-term photovoltaic power generation forecasting based on random forest feature selection and CEEMD: A case study. Applied Soft Computing. 93: 106389. doi: https://doi.org/10.1016/j.asoc.2020.106389.
24. [24] Gwetu, M.V., Tapamo, J.-R., Viriri, S., (2019). Exploring the Impact of Purity Gap Gain on the Efficiency and Effectiveness of Random Forest Feature Selection BT - Computational Collective Intelligence. In: Nguyen, N.T., Chbeir, R., Exposito, E., Aniorté, P., Trawiński, B., editors. Cham: Springer International Publishing p. 340–52.
25. [25] Li, X., Wang, Y., Basu, S., Kumbier, K., Yu, B., (2019). A debiased MDI feature importance measure for random forests. Advances in Neural Information Processing Systems. 32(NeurIPS).
26. [26] Cui, W., Sun, Z., Ma, H., Wu, S., (2020). The Correlation Analysis of Atmospheric Model Accuracy Based on the Pearson Correlation Criterion. IOP Conference Series: Materials Science and Engineering. 780(3). doi: 10.1088/1757-899X/780/3/032045.
27. [27] Cowen-Rivers, A.I., Lyu, W., Tutunov, R., Wang, Z., Grosnit, A., Rhys, R., et al., (2022). HEBO: Pushing The Limits of Sample-Efficient Hyperparameter Optimisation. Journal of Artificial Intelligence Research. 74(February): 1269–349. doi: 10.1613/JAIR.1.13643.
28. [28] Kervanci, I.S., Akay, M.F., Özceylan, E., (2024). Bitcoin price prediction using LSTM, GRU and hybrid LSTM-GRU with bayesian optimization, random search, and grid search for the next days. Journal of Industrial and Management Optimization. 20(2): 570–88. doi: 10.3934/jimo.2023091.
29. [29] Jiang, S.Y., Wang, L.X., (2016). Efficient feature selection based on correlation measure between continuous and discrete features. Information Processing Letters. 116(2): 203–15. doi: 10.1016/j.ipl.2015.07.005.
30. [30] Liu, Y., Zou, X., Ma, S., Avdeev, M., Shi, S., (2022). Feature selection method reducing correlations among features by embedding domain knowledge. Acta Materialia. 238: 118195. doi: 10.1016/j.actamat.2022.118195.
31. [31] Zhang, F., O’Donnell, L.J., (2019). Support vector regression. Elsevier Inc.