OPTIMIZING RANDOM FOREST REGRESSOR PERFORMANCE IN BEEF QUALITY PREDICTION THROUGH HYPERPARAMETER TUNING

• Dr. Paige L. Bennett

  Department of Journalism and Mass Communication, University of Georgia, Athens, GA, USA

  Author

• Dr. Aaron J. Myers

  Department of Political Psychology, University of Nebraska–Lincoln, Lincoln, NE, USA

  Author

The continuous escalation in global meat consumption, particularly beef, underscores the critical need for sophisticated and efficient methodologies to ascertain and predict meat quality attributes. Traditional approaches to quality assessment, often reliant on invasive and laborious laboratory analyses, present significant logistical and economic challenges within the modern meat industry. In response to these limitations, the integration of advanced machine learning paradigms, specifically ensemble learning models such as the Random Forest Regressor (RFR), has emerged as a highly promising avenue for developing rapid, non-destructive, and precise prediction models. However, the inherent complexity of these models implies that their predictive efficacy is profoundly contingent upon the meticulous configuration of their intrinsic hyperparameters. This comprehensive investigation meticulously explores the intricate relationship between hyperparameter tuning and the predictive accuracy of an RFR model specifically tailored for beef quality assessment.

Leveraging a robust dataset encompassing a diverse array of physico-chemical parameters and spectral signatures obtained via Near-Infrared (NIR) spectroscopy, a rigorous and systematic hyperparameter optimization strategy was implemented. This strategy prominently featured Randomized Search Cross-Validation, a statistically efficient technique designed to traverse a wide parameter space. The specific hyperparameters subjected to optimization included the number of constituent trees (n_estimators), the criteria for feature selection at each node split (max_features), the minimum number of data samples mandated for a leaf node (min_samples_leaf), and the requisite minimum samples for an internal node to undergo splitting (min_samples_split). The empirical findings unequivocally demonstrate a marked enhancement in the RFR model's overall predictive performance post-tuning. This improvement is robustly evidenced by substantial increases in the Coefficient of Determination (R2) values across all predicted beef quality attributes, complemented by significant reductions in error metrics such as Mean Absolute Error (MAE), Mean Squared Error (MSE), and Root Mean Squared Error (RMSE). The average R2 improvement across all parameters was approximately 14% compared to models using default parameters. These compelling results accentuate the indispensable role of comprehensive hyperparameter optimization in cultivating high-performing, resilient, and accurate predictive models for the multifaceted domain of beef quality assessment.

[1] M. Lima, R. Costa, I. Rodrigues, J. Lameiras, and G. Botelho, “A narrative review of alternative protein sources: highlights on meat, fish, egg and dairy analogues,” Foods, vol. 11, no. 14, 2022, doi: 10.3390/foods11142053.

[2] M. Molfetta et al., “Protein sources alternative to meat: state of the art and involvement of fermentation,” Foods, vol. 11, no. 14, 2022, doi: 10.3390/foods11142065.

[3] B. Fletcher et al., “Advances in meat spoilage detection: a short focus on rapid methods and technologies,” CYTA-Journal of Food, vol. 16, no. 1, pp. 1037–1044, 2018, doi: 10.1080/19476337.2018.1525432.

[4] V. Tesson, M. Federighi, E. Cummins, J. de O. Mota, S. Guillou, and G. Boué, “A systematic review of beef meat quantitative microbial risk assessment models,” International Journal of Environmental Research and Public Health, vol. 17, no. 3, 2020, doi: 10.3390/ijerph17030688.

[5] W. Barragán-Hernández, L. Mahecha-Ledesma, J. Angulo-Arizala, and M. Olivera-Angel, “Near-infrared spectroscopy as a beef quality tool to predict consumer acceptance,” Foods, vol. 9, no. 8, 2020, doi: 10.3390/foods9080984.

[6] G. Ripoll et al., “Near-infrared reflectance spectroscopy for predicting the phospholipid fraction and the total fatty acid composition of freeze-dried beef,” Sensors, vol. 21, no. 12, 2021, doi: 10.3390/s21124230.

[7] A. Sahar et al., “Online prediction of physico-chemical quality attributes of beef using visible-near-infrared spectroscopy and chemometrics,” Foods, vol. 8, no. 11, 2019, doi: 10.3390/foods8110525.

[8] N. Patel, H. Toledo-Alvarado, A. Cecchinato, and G. Bittante, “Predicting the content of 20 minerals in beef by different portable near-infrared (NIR) spectrometers,” Foods, vol. 9, no. 10, 2020, doi: 10.3390/foods9101389.

[9] S. Savoia et al., “Prediction of meat quality traits in the abattoir using portable and hand-held near-infrared spectrometers,” Meat Science, vol. 161, 2020, doi: 10.1016/j.meatsci.2019.108017.

[10] S. Savoia, A. Albera, A. Brugiapaglia, L. Di Stasio, A. Cecchinato, and G. Bittante, “Prediction of meat quality traits in the abattoir using portable near-infrared spectrometers: heritability of predicted traits and genetic correlations with laboratory-measured traits,” Journal of Animal Science and Biotechnology, vol. 12, no. 1, 2021, doi: 10.1186/s40104-021-00555-5.

[11] I. M. N. Perez, L. J. P. Cruz-Tirado, A. T. Badaró, M. M. de Oliveira, and D. F. Barbin, “Present and future of portable/handheld near-infrared spectroscopy in chicken meat industry,” NIR news, vol. 30, no. 5–6, pp. 26–29, 2019, doi: 10.1177/0960336019861476.

[12] M. Simoni, A. Goi, M. De Marchi, and F. Righi, “The use of visible/near-infrared spectroscopy to predict fibre fractions, fibre-bound nitrogen and total-tract apparent nutrients digestibility in beef cattle diets and faeces,” Italian Journal of Animal Science, vol. 20, no. 1, pp. 814–825, 2021, doi: 10.1080/1828051X.2021.1924884.

[13] C. N. Sánchez, M. T. Orvañanos-Guerrero, J. Domínguez-Soberanes, and Y. M. Álvarez-Cisneros, “Analysis of beef quality according to color changes using computer vision and white-box machine learning techniques,” Heliyon, vol. 9, no. 7, 2023, doi: 10.1016/j.heliyon.2023.e17976.

[14] T. Qiao, J. Ren, C. Craigie, J. Zabalza, C. Maltin, and S. Marshall, “Quantitative prediction of beef quality using visible and NIR spectroscopy with large data samples under industry conditions,” Journal of Applied Spectroscopy, vol. 82, no. 1, pp. 137–144, 2015, doi: 10.1007/s10812-015-0076-1.

[15] G. Biau and E. Scornet, “A random forest guided tour,” Test, vol. 25, no. 2, pp. 197–227, 2016, doi: 10.1007/s11749-016-0481-7.

[16] H. Pu, J. Yu, D. W. Sun, Q. Wei, X. Shen, and Z. Wang, “Distinguishing fresh and frozen-thawed beef using hyperspectral imaging technology combined with convolutional neural networks,” Microchemical Journal, vol. 189, 2023, doi: 10.1016/j.microc.2023.108559.

[17] R. Kasarda, N. Moravčíková, G. Mészáros, M. Simčič, and D. Zaborski, “Classification of cattle breeds based on the random forest approach,” Livestock Science, vol. 267, 2023, doi: 10.1016/j.livsci.2022.105143.

[18] Y. Lin, J. Ma, D. W. Sun, J. H. Cheng, and Q. Wang, “A pH-Responsive colourimetric sensor array based on machine learning for real-time monitoring of beef freshness,” Food Control, vol. 150, 2023, doi: 10.1016/j.foodcont.2023.109729.

[19] F. Pedregosa et al., “Scikit-learn: machine learning in Python,” Journal of Machine Learning Research, vol. 12, pp. 2825–2830, 2011.

[20] P. Probst, M. N. Wright, and A. L. Boulesteix, “Hyperparameters and tuning strategies for random forest,” Wiley Interdisciplinary Reviews: Data Mining and Knowledge Discovery, vol. 9, no. 3, 2019, doi: 10.1002/widm.1301.

[21] S. Bernard, L. Heutte, and S. Adam, “Influence of hyperparameters on random forest accuracy,” in Multiple Classifier Systems, Berlin, Heidelberg: Springer, 2009, pp. 171–180, doi: 10.1007/978-3-642-02326-2_18.

[22] Scikit Learn, “RandomizedSearchCV,” Scikit Learn. Accessed: May 17, 2024. [Online]. Available: https://scikit-learn.org/stable/modules/generated/sklearn.model_selection.RandomizedSearchCV.html.

[23] R. Raafi’udin, Y. A. Purwanto, I. S. Sitanggang, and D. A. Astuti, “Feature selection model development on near-infrared spectroscopy data,” International Journal of Advanced Computer Science and Applications, vol. 15, no. 1, 2024, doi: 10.14569/ijacsa.2024.0150163.

[24] Scikit Learn, “RandomForestRegressor,” Scikit Learn. Accessed: May 17, 2024. [Online]. Available: https://scikit-learn.org/stable/modules/generated/sklearn.ensemble.RandomForestRegressor.html.

[25] D. Chicco, M. J. Warrens, and G. Jurman, “The coefficient of determination R-squared is more informative than SMAPE, MAE, MAPE, MSE and RMSE in regression analysis evaluation,” PeerJ Computer Science, vol. 7, pp. 1–24, 2021, doi: 10.7717/PEERJ-CS.623.

[26] A. Chugh, “MAE, MSE, RMSE, coefficient of determination, adjusted R squared—which metric is better,” Medium, 2020. Accessed: Nov 11, 2024. [Online]. Available: https://medium.com/analytics-vidhya/mae-mse-rmse-coefficient-of-determination-adjusted-r-squared-which-metric-is-better-cd0326a5697e.

• pdf

All articles published by The Parthenon Frontiers and its associated journals are distributed under the terms of the Creative Commons Attribution (CC BY 4.0) International License unless otherwise stated. 

Authors retain full copyright of their published work. By submitting their manuscript, authors agree to grant The Parthenon Frontiers a non-exclusive license to publish, archive, and distribute the article worldwide. Authors are free to:

• Share their article on personal websites, institutional repositories, or social media platforms.

• Reuse their content in future works, presentations, or educational materials, provided proper citation of the original publication.