Comparison of Different Machine Learning Methods for Estimating Agricultural Products

Yıl 2022, Cilt: 3 Sayı: 1, 12 - 21, 29.06.2022

Halit Çetiner

Öz

Dünya ekonomisi üzerinde farklı etkileri olan mahsullerin verim tahminlerini gerçekleştirmek için iki farklı makine öğrenmesi yöntemi ile regresyon analizi yapılmıştır.Regresyon analizinde kullanılan iki yöntem DVM (Destek Vektör Makinesi) ve HTGA (Histogram Tabanlı Gradyan Artırma) regresyon analiz yöntemleridir. Her iki regresyon analiz yöntemi de çok farklı tarımsal problemlerin temelinde yatan verim tahminini en az hata ile bulmaya çalışmaktadır. Bu anlamda deneysel çalışmalar gerçekleştirebilmek için 1961 yılından başlayarak 2016 yılına kadar ki 35 yıllık aralıkta Dünya Veri Bankası ve FAO dünya tarım örgütü veri tabanlarında bulunan yağış, sıcaklık, pestisit girdi değerleri kullanarak verim tahmininde bulunulmuştur. Verim tahminleri sonucunda HTGA ile 94% R2 puanına ulaşılırken, DVM Poly çekirdeği ile 91% R2 puanına erişilmiştir. DVM yönteminde ise 3 farklı çekirdek kullanarak regresyon analizi gerçekleştirilmiştir. Poly çekirdek değeri sonuçları 91% R2 puanına ulaşırken RBF ve Linear çekirdek değerleri sırasıyla 81% ve 69% R2 puanına ulaşmıştır.

Anahtar Kelimeler

DVM, HTGA, Verim tahmini, Regresyon analizi

Comparison of Different Machine Learning Methods for Estimating Agricultural Products

Öz

Regression analysis was carried out with two different machine learning methods in order to realize the yield estimates of crops that have different effects on the world economy. The two methods used in regression analysis are SVM (Support Vector Machine) and HTGA (Histogram Based Gradient Augmentation) regression analysis methods. Both regression analysis methods try to find the yield estimation underlying many different agricultural problems with the least error. In this sense, to carry out experimental studies, yield estimations were made using precipitation, temperature, pesticide input values found in the World Data Bank and FAO World Agricultural Organization databases in the 35-year interval from 1961 to 2016. As a result of efficiency estimation, the 94% R2 score was reached with HTGA, while the 91% R2 score was reached with the SVM Poly core. In the SVM method, regression analysis was performed using 3 different kernels. The results of the Poly core values reached 91% R2 scores, while the RBF and linear core values reached 81% and 69% R2 scores, respectively.

Anahtar Kelimeler

SVM, HTGA, Yield estimation, Regression analysis

Kaynakça

• [1] Ö. Vanli, B. B. Ustundag, I. Ahmad, I. M. Hernandez-Ochoa, and G. Hoogenboom, “Using crop modeling to evaluate the impacts of climate change on wheat in southeastern turkey,” Environ. Sci. Pollut. Res., vol. 26, no. 28, pp. 29397–29408, 2019, doi: 10.1007/s11356-019-06061-6.
• [2] W. Nasim et al., “Future risk assessment by estimating historical heat wave trends with projected heat accumulation using SimCLIM climate model in Pakistan,” Atmos. Res., vol. 205, pp. 118–133, 2018, doi: https://doi.org/10.1016/j.atmosres.2018.01.009.
• [3] I. Ahmad et al., “Yield Forecasting of Spring Maize Using Remote Sensing and Crop Modeling in Faisalabad-Punjab Pakistan,” J. Indian Soc. Remote Sens., vol. 46, no. 10, pp. 1701–1711, 2018, doi: 10.1007/s12524-018-0825-8.
• [4] J. Cao et al., “Wheat yield predictions at a county and field scale with deep learning, machine learning, and google earth engine,” Eur. J. Agron., vol. 123, p. 126204, 2021, doi: https://doi.org/10.1016/j.eja.2020.126204.
• [5] I. FAO, “WFP (2015). The state of food insecurity in the World. Meeting the 2015 international hunger targets: taking stock of uneven progress. Rome, FAO.” 2017.
• [6] F. Dodds and J. Bartram, The water, food, energy and climate Nexus: Challenges and an agenda for action. Routledge, 2016.
• [7] R. Patel, “Dataset,” 2022. https://www.kaggle.com/patelris/crop-yield-prediction-dataset
• [8] O. O. Abayomi-Alli, R. Damaševičius, S. Misra, and R. Maskeliūnas, “Cassava disease recognition from low-quality images using enhanced data augmentation model and deep learning,” Expert Syst., vol. 38, no. 7, p. e12746, Nov. 2021, doi: https://doi.org/10.1111/exsy.12746.
• [9] M. B. Cole, M. A. Augustin, M. J. Robertson, and J. M. Manners, “The science of food security,” npj Sci. Food, vol. 2, no. 1, pp. 1–8, 2018.
• [10] T. Stevens and K. Madani, “Future climate impacts on maize farming and food security in Malawi,” Sci. Rep., vol. 6, no. 1, pp. 1–14, 2016.
• [11] D. Tilman, C. Balzer, J. Hill, and B. L. Befort, “Global food demand and the sustainable intensification of agriculture,” Proc. Natl. Acad. Sci., vol. 108, no. 50, pp. 20260–20264, 2011.
• [12] A. F. Al-daour and M. O. Al-shawwa, “Classification of Banana Fruits Using Deep Learning,” 2020.
• [13] K. A. Beals, “Potatoes, nutrition and health,” Am. J. Potato Res., vol. 96, no. 2, pp. 102–110, 2019.
• [14] Y. Zhu and M. Ghosh, “Temperature control, emission abatement and costs: key EMF 27 results from Environment Canada’s Integrated Assessment Model,” Clim. Change, vol. 123, no. 3, pp. 571–582, 2014.
• [15] C. Gonzalo-Martín, A. García-Pedrero, and M. Lillo-Saavedra, “Improving deep learning sorghum head detection through test time augmentation,” Comput. Electron. Agric., vol. 186, p. 106179, 2021, doi: https://doi.org/10.1016/j.compag.2021.106179.
• [16] C. W. Mundia, S. Secchi, K. Akamani, and G. Wang, “A Regional Comparison of Factors Affecting Global Sorghum Production: The Case of North America, Asia and Africa’s Sahel,” Sustainability , vol. 11, no. 7. 2019. doi: 10.3390/su11072135.
• [17] Z. Lin and W. Guo, “Sorghum Panicle Detection and Counting Using Unmanned Aerial System Images and Deep Learning,” Front. Plant Sci., vol. 11, 2020, doi: 10.3389/fpls.2020.534853.
• [18] R. Battisti, P. C. Sentelhas, and K. J. Boote, “Inter-comparison of performance of soybean crop simulation models and their ensemble in southern Brazil,” F. Crop. Res., vol. 200, pp. 28–37, 2017, doi: https://doi.org/10.1016/j.fcr.2016.10.004.
• [19] P. C. Sentelhas, R. Battisti, G. M. S. Câmara, J. R. B. Farias, A. C. Hampf, and C. Nendel, “The soybean yield gap in Brazil–magnitude, causes and possible solutions for sustainable production,” J. Agric. Sci., vol. 153, no. 8, pp. 1394–1411, 2015.
• [20] D. R. Rankine et al., “Parameterizing the FAO AquaCrop Model for Rainfed and Irrigated Field-Grown Sweet Potato,” Agron. J., vol. 107, no. 1, pp. 375–387, Jan. 2015, doi: https://doi.org/10.2134/agronj14.0287.
• [21] N. Verter and V. Bečvářová, “An analysis of yam production in Nigeria,” Acta Univ. Agric. Silvic. Mendelianae Brun., vol. 63, no. 2, pp. 659–665, 2015.
• [22] Y. Cai et al., “Integrating satellite and climate data to predict wheat yield in Australia using machine learning approaches,” Agric. For. Meteorol., vol. 274, pp. 144–159, 2019, doi: https://doi.org/10.1016/j.agrformet.2019.03.010.
• [23] R. Alvarez, “Predicting average regional yield and production of wheat in the Argentine Pampas by an artificial neural network approach,” Eur. J. Agron., vol. 30, no. 2, pp. 70–77, 2009, doi: https://doi.org/10.1016/j.eja.2008.07.005.
• [24] A. B. Potgieter, D. B. Lobell, G. L. Hammer, D. R. Jordan, P. Davis, and J. Brider, “Yield trends under varying environmental conditions for sorghum and wheat across Australia,” Agric. For. Meteorol., vol. 228–229, pp. 276–285, 2016, doi: https://doi.org/10.1016/j.agrformet.2016.07.004.
• [25] Y. Li, “Prediction of energy consumption: Variable regression or time series? A case in China,” Energy Sci. Eng., vol. 7, no. 6, pp. 2510–2518, Dec. 2019, doi: 10.1002/ese3.439.
• [26] J. A. Mathieu and F. Aires, “Assessment of the agro-climatic indices to improve crop yield forecasting,” Agric. For. Meteorol., vol. 253–254, pp. 15–30, 2018, doi: https://doi.org/10.1016/j.agrformet.2018.01.031.
• [27] J. Zhang, Y. Chen, and Z. Zhang, “A remote sensing-based scheme to improve regional crop model calibration at sub-model component level,” Agric. Syst., vol. 181, p. 102814, 2020, doi: https://doi.org/10.1016/j.agsy.2020.102814.
• [28] G. Azzari, M. Jain, and D. B. Lobell, “Towards fine resolution global maps of crop yields: Testing multiple methods and satellites in three countries,” Remote Sens. Environ., vol. 202, pp. 129–141, 2017.
• [29] Z. Jin, G. Azzari, M. Burke, S. Aston, and D. B. Lobell, “Mapping smallholder yield heterogeneity at multiple scales in Eastern Africa,” Remote Sens., vol. 9, no. 9, p. 931, 2017.
• [30] A. Kern et al., “Statistical modelling of crop yield in Central Europe using climate data and remote sensing vegetation indices,” Agric. For. Meteorol., vol. 260, pp. 300–320, 2018.
• [31] K. Kuwata and R. Shibasaki, “Estimating crop yields with deep learning and remotely sensed data,” in 2015 IEEE International Geoscience and Remote Sensing Symposium (IGARSS), 2015, pp. 858–861.
• [32] D. Khanna, R. Sahu, V. Baths, and B. Deshpande, “Comparative Study of Classification Techniques (SVM, Logistic Regression and Neural Networks) to Predict the Prevalence of Heart Disease,” Int. J. Mach. Learn. Comput., vol. 5, pp. 414–419, Oct. 2015, doi: 10.7763/IJMLC.2015.V5.544.
• [33] G. Ke et al., “Lightgbm: A highly efficient gradient boosting decision tree,” Adv. Neural Inf. Process. Syst., vol. 30, 2017.
• [34] D. C. Montgomery, E. A. Peck, and G. G. Vining, Introduction to linear regression analysis. John Wiley & Sons, 2021.
• [35] D. W. Hosmer Jr, S. Lemeshow, and R. X. Sturdivant, Applied logistic regression, vol. 398. John Wiley & Sons, 2013.