Garlic Fresh Weight Prediction using Machine Learning

Min-Seo Kang; Hyo-In Yoon; Hye-Jin Lee; Jin-Hee Kim; Tae-Bok Kim; Tae-Chul Seo; Seung-Hwan Wi

doi:10.12791/KSBEC.2025.34.1.010

All Issue

2025 Vol.34, Issue 1 Preview Page Next Page

Original Articles

Garlic Fresh Weight Prediction using Machine Learning 기계학습을 활용한 마늘 생체중 예측

31 January 2025. pp. 10-18

PDF XML

Abstract

This study aimed to develop a model for predicting garlic fresh weight using various machine learning techniques based on growth data, including leaf length, leaf count, and growing degree days as meteorological data. In the validation dataset, the Random forest model exhibited the highest performance with R² of 0.924, RMSE (g) of 13.583, and MAE of 8.885. In the test dataset, the Catboost model outperformed others with R² of 0.928, RMSE (g) of 13.486, and MAE of 9.181. However, a Weighted ensemble model trained with weights of 0.5, 0.3, and 0.2 for Catboost, Random forest, and LightGBM models, respectively, demonstrated the second-best performance in both the validation and test datasets, with R² of 0.922 and 0.923, RMSE (g) of 13.752 and 13.992, and MAE of 8.877 and 9.437, respectively. Based on these findings, the Weighted ensemble model was deemed the optimal choice. Thus, it is suggested that farmers can employ machine learning techniques to predict garlic fresh weight using only phenotypic and meteorological data, facilitating field monitoring. However, further data acquisition and validation over multiple years are essential to refine the model’s performance.

Keywords

Garlic

Machine Learning

Fresh weight

본 연구는 표현 형질 생육 데이터인 엽장, 엽 수와 기상 데이터인 생육도일을 활용하여 여러 기계 학습을 통해 마늘의 생체중을 예측하는 모델을 개발하고자 하였다. 검증 데이터에서 random forest 모델의 결정계수가 0.924, 평균제곱근오차(g)는 13.583 그리고 평균절대오차는 8.885로 가장 우수하였다. 평가 데이터에서는 Catboost 모델이 결정계수가 0.928, 평균제곱근오차(g)는 13.486 그리고 평균절대오차는 9.181로 가장 우수하였다. 그러나 Catboost, Random forest 그리고 LightGBM 모델을 0.5, 0.3 그리고 0.2 가중치를 두어 학습한 Weighted ensemble 모델이 마늘 생체중 예측의 검증 및 평가에 있어서 검증 데이터의 결정계수가 0.922, 평균제곱근오차(g)가 13.752 그리고 평균절대오차는 8.877이었으며 평가 데이터에서는 결정계수가 0.923, 평균제곱근오차(g)가 13.992 그리고 평균절대오차가 9.437로 두 번째로 우수한 결과를 나타내었다. 이러한 결과들을 종합적으로 미루어 보았을 때, Weighted ensemble 모델이 모델의 안정성 측면에서 최적의 모델이라고 판단하였다. 따라서 농가들이 표현 형질과 기상 데이터만으로도 기계학습 기법을 통하여 마늘의 생체중 예측을 통해 작형 모니터링이 가능할 것으로 보이며 추가적으로 다년도 데이터 취득과 검증을 통하여 성능을 고도화가 가능할 것으로 판단된다.

키워드

마늘

머신러닝

생체중

References

Breiman L. 2001, Random forests. Mach Learn 45:5-32. doi:10.1023/A:1010950718922

10.1023/A:1010950718922

Chen T., and C. Guestrin 2016, Xgboost: A scalable tree boosting system. In Proceedings of the 22nd acm sigkdd international conference on knowledge discovery and data mining, pp 785-794.

10.1145/2939672.2939785

El-Zohiri S.S.M., and A.A. Farag 2014, Relation planting date, cultivars and growing degree-days on growth, yield and quality of garlic. Middle East J Agri Res 3:1169-1183.

Hoerl A.E., and R.W Kennard 1970, Ridge regression:biased estimation for nonrthogonal problems. Techonmetrics 12:55-67. doi:10.1080/00401706.1970.10488634

10.1080/00401706.1970.10488634

Jang S.H., C.S. Ryu, Y.S Kang, S.R. Jun, J.W. Park, H.Y. Song, K.S. Kang, D.W. Kang, K. Zou, and T.H Jun 2019, Estimation of fresh weight, dry weight, and leaf area index of soybean plant using multispectral camera mounted on rotor-wing uav. Korean J Agric For Meteorol 21:327-336. (in Korean) doi:10.5532/KJAFM.2019.21.4.327

Jo J.S., H.S. Sim, S.B. Jung, Y.H Moon, W.J Jo, U.J Woo, and S.K. Kim 2022, Estimation and validation of the leaf areas of five June-bearing strawberry(fragaria × ananassa) cultivars using non-destructive methods. J Bio-Env Con 31:98-103. (in Korean) doi:10.12791/KSBEC.2022.31.2.098

10.12791/KSBEC.2022.31.2.098

Jung D.H., Y.S. Chung, and H.S. Hwang 2023, Development of a model for estimating leaf area and the number of flower using leaf length and width of farfugium japonicum kitam. J Bio-Env Con 32:115-121. (in Korean) doi:10.12791/KSBEC.2022.32.2.115

10.12791/KSBEC.2023.32.2.115

Kang M.S., J.S. Shim, H.J. Lee, H.J. Lee, Y.A. Jang, W.M. Lee, S.G. Lee, and S.H. Wi 2023, Development of kimchi cabbage growth prediction models based on image and temperature data. J Bio-Env Con 32:366-376. (in Korean) doi:10.12791/KSBEC.2023.32.4.366

10.12791/KSBEC.2023.32.4.366

Ke G., Q. Meng, T. Finley, T. Wang, W. Chen, W. Ma, Q. Ye, and T.Y. Liu 2017, Lightgbm: A highly efficient gradient boosting decision tree. Advances in neural information processing systems, 30. NeurIPS doi:10.1109/ICCSE.2019.8845529

10.1109/ICCSE.2019.8845529

Kim N.E., H.S. Han, E. Arulmozhi, B.E. Moon, Y.W Choi, and H.T. Kim 2022, Prediction of greenhouse strawberry production using machine learning algorithm. J Bio-Env Con 31:1-7. (in Korean) doi:10.12791/KSBEC.2022.31.1.001

10.12791/KSBEC.2022.31.1.001

KOSTAT 2023. Survey of garlic production 2023

KREI 2023. Agricultural outlook 2023 Korea. 3-38. (in Korean)

Lee C.J., C.S. Park, J.S. Kim, and J.G. Baek 2015, A study on improving classification performance for manufacturing process data with multicollinearity and imbalanced distribution. J Korean Institute of Industrial Engineers 41:25-33. (in Korean) doi:10.7232/JKIIE.2015.41.1.025

10.7232/JKIIE.2015.41.1.025

Lee I.J., G.W Park, S.Y. Choi, C.K. Yun, J.S. Lee, K.H Kim, K.S. Ahn, S.Y. Nam, and T.J. Kim 2013, Effects of the time sowing on growth and yield of northern type garlic. Korean J Intl Agri 25:260-264. (in Korean) doi:10.12719/KSIA.2013.25.3.260

10.12719/KSIA.2013.25.3.260

Lee J.M., J.Y. Jeong, and H.G. Choi 2022, Estimation of leaf area using leaf length, leaf width, and lamina length in tomato. J Bio-Env Con 31:325-331. (in Korean) doi:10.12791/KSBEC.2022.31.4.325

10.12791/KSBEC.2022.31.4.325

McMaster G.S., and W.W. Wilhelm 1997, Growing degree-days: one equation, two interpretations. Agric For Meteorol 87:291-300. doi:10.1016/S0168-1923(97)00027-0

10.1016/S0168-1923(97)00027-0

Prokhorenkova L., G. Gusev, A. Vorobev, A.V. Dorogush, and A. Gulin 2018, CatBoost: unbiased boosting with categorical features. Advances in neural information processing systems, 31. doi:10.48550/arXiv.1706.09516

Tibshirani R. 1996, Regression shrinkage and selection via the lass. J R Statist Soc B 58:267-288. doi:10.1111/j.2517-6161.1996.tb02080.x

10.1111/j.2517-6161.1996.tb02080.x

Wi S.H., H.J. Lee, I.H. Yu, Y. Jang, K.H. Yeo, S. An, and J.H. Lee 2020, Analysis of effect of environment on growth and yield of autumn kimchi cabbage in Jeonnam province using big data. Korean J Agric For Meteorol 22:183-193. (in Korean) doi:10.5532/KJAFM.2020.22.3.183

Wi S.H., K.H. Moon, E.Y. Song, I.C. Son, S.J Oh, and Y.Y Cho 2017, Growth and fresh bulb weight model in harvest time of southern type garlic var. 'Namdo' based on temperature. Protected Horti and Plant Factory 26:13-18. (in Korean) doi:10.12791/KSBEC.2017.26.1.13

10.12791/KSBEC.2017.26.1.13

Wolpert D.H. 1992, Stacked generalization. Neural networks, 5:241-259. doi:10.1016/S0893-6080(05)80023-1

10.1016/S0893-6080(05)80023-1

Zhang N., X. Zhou, M. Kang, B.G. Hu, E. Heuvelink and L.F Marcelis 2022, Machine learning versus crop growth models: an ally, not a rival. AoB Plants 15:1-7. doi:10.1093/aobpla/plac061

10.1093/aobpla/plac06136751366PMC9893870

Zhou Z.H. 2014, Ensemble methods. Combining pattern classifiers, Wiley, Hoboken pp 186-229.

10.1002/9781118914564.ch6

Information

Publisher :The Korean Society for Bio-Environment Control
Publisher(Ko) :(사)한국생물환경조절학회
Journal Title :Journal of Bio-Environment Control
Journal Title(Ko) :생물환경조절학회지
Volume : 34
No :1
Pages :10-18
Received Date : 2024-09-27
Revised Date : 2024-11-04
Accepted Date : 2024-11-11
DOI :https://doi.org/10.12791/KSBEC.2025.34.1.010

Journal of Bio-Environment Control ISSN:1229-4675(Print) 2765-3641(Online) 생물환경조절학회지

All Issue