RSM vs Machine Learning for moisture prediction in the convective drying of avocado pulp: Insights from a 30-Run CCD Datasets

Awang Bono; Zykamilia Kamin; Dona Stacy Petrus; Muhamad Afif Naqiudien Aladin; Mohd Hardyianto Vai Bahrun.

The amount of moisture content in drying process is crucial. Determining accurate process control, product quality and energy efficiency in food processing is necessary to ensure the quality of the drying food. In this study, Response Surface Methodology (RSM), a feedforward Neural Network (NN), and a Random Forest (RF) model were developed for moisture content of avocado pulp dried under controlled convective conditions. A Central Composite Design with number of experiments = 30 was used to explore four factors which are hot-air temperature (denoted as A, ranging from 40 to 70 °C), drying time (B, 13–20 h), wind speed (C, 0.12–0.26 m/s), and raw material thickness (D, 0.5–1.25 cm). A second-order polynomial model which was developed using RSM, was fitted to the experimental data and compared with other models, NN and RF using cross-validation to improve robustness. The results showed that RSM model performed significantly better than the other two models. The RSM model achieved coefficient of determination of 0.80 followed by NN, 0.43 and RF, 0.34. Feature-importance was used on ML models to determine the ranking of the factors based on how significant the factors influence the drying process. It was identified that wind speed and drying time as the main factors to represent the final moisture content, directly affect mass-transfer control in this application. The results highlight that small datasets can outperform the machine learning model especially for characterizing food drying processes.

KEYWORDS: Drying kinetics; Response Surface Methodology; Neural Network; Random Forest; Moisture prediction



Download this PDF file

REFERENCES

1. Aksoy, A., Karasu, S., Akcicek, A. & Kayacan, S. 2019. Effects of different drying methods on drying kinetics, microstructure, color, and the rehydration ratio of minced meat. Foods, 8(6), 216.
2. Başlar, M., Kılıçlı, M. & Yalınkılıç, B. 2015. Dehydration kinetics of salmon and trout fillets using ultrasonic vacuum drying as a novel technique. Ultrasonics Sonochemistry, 27, 495–502.
3. Delfiya, D. S. A., Prashob, K., Murali, S., Alfiya, P. V, Samuel, M. P. & Pandiselvam, R. 2022. Drying kinetics of food materials in infrared radiation drying: A review. Journal of Food Process Engineering, 45(6), e13810.
4. Fan, L., Pei, Y., Zhang, L., Kong, J. & Xu, W. 2025. Applications of Machine Learning Models in Agricultural Product Drying: A Comprehensive Review of Advances, Challenges, and Prospects. Food and Bioprocess Technology, 18, 10047–10085.
5. Golpour, I., Kaveh, M., Amiri Chayjan, R. & Guiné, R. P. F. 2020. Optimization of infrared-convective drying of white mulberry fruit using response surface methodology and development of a predictive model through artificial neural network. International Journal of Fruit Science, 20(sup2), S1015–S1035.
6. Khan, M. I. H., Sablani, S. S., Joardder, M. U. H. & Karim, M. A. 2022. Application of machine learning-based approach in food drying: Opportunities and challenges. Drying Technology, 40(6), 1051–1067.
7. Kowarit, S., Sathapornprasath, K., & Jansri, S. N. 2024. Application of hot air-derived RSM conditions and shading for solar drying of avocado pulp and its properties. Solar Energy, 278, 112768.
8. Kuhn, M. & Johnson, K. 2019. Feature engineering and selection: A practical approach for predictive models. Chapman and Hall/CRC.
9. Myers, R. H., Montgomery, D. C. & Anderson-Cook, C. M. 2016. Response Surface Methodology: Process and Product Optimization Using Designed Experiments (4th Ed). Wiley.
10. Pedregosa, F., Varoquaux, G., Gramfort, A., Michel, V., Thirion, B., Grisel, O., Blondel, M., Prettenhofer, P., Weiss, R. & Dubourg, V. 2011. Scikit-learn: Machine learning in Python. The Journal of Machine Learning Research, 12, 2825–2830.
11. Przybył, K. & Koszela, K. 2023. Applications MLP and other methods in artificial intelligence of fruit and vegetable in convective and spray drying. Applied Sciences, 13(5), 2965.
12. Rather, I. H., Kumar, S. & Gandomi, A. H. 2024. Breaking the data barrier: a review of deep learning techniques for democratizing AI with small datasets. Artificial Intelligence Review, 57(9), 226.
13. Refaeilzadeh, P., Tang, L. & Liu, H. 2018. Cross Validation. In: Liu, L. & Özsu, M.T. (eds). Encyclopedia of Database Systems. Springer. pp. 532–538.
14. Zantvoort, K., Nacke, B., Görlich, D., Hornstein, S., Jacobi, C. & Funk, B. 2024. Estimation of minimal data sets sizes for machine learning predictions in digital mental health interventions. npj Digital Medicine, 7(1), 361.