Machine Learning for Soil Attribute Prediction: An Effectiveness and Dimensionality Reduction Analysis

  • José Solenir L. Figuerêdo UEFS
  • Marcos Eduardo de C. Ferreira UEFS
  • Rodrigo T. Calumby UEFS

Resumo


Traditional soil fertility analyzes are laborious, expensive, timeconsuming and produce hazardous waste. Although many works using machine learning (ML) has been done to address these issues, some algorithms and dimensionality reduction strategies require further investigation. Therefore, in this study we evaluated the potential of Support Vector Regression and Ridge regression in determining soil attributes, and compared principal components regression and partial least squares regression (PLSR). The results showed that Ridge was the most effective model. In addition, our experiments revealed that PLSR was able to achieve statistically equivalent results, and in some cases superior to the baseline, but using a much smaller average number of components.

Referências

Benedet, L. et al. (2021). Rapid soil fertility prediction using x-ray fluorescence data and machine learning algorithms. Catena, 197:105003.

Folorunso, O. et al. (2023). Exploring machine learning models for soil nutrient properties prediction: A systematic review. Big Data and Cognitive Computing, 7(2).

Laili, A. R. et al. (2020). Prediction of soil macronutrient (nitrate and phosphorus) using near-infrared (NIR) spectroscopy and machine learning. AIP Conference Proceedings, 2203(1):020061.

Liu, C. et al. (2022). Partial least squares regression and principal component analysis: similarity and differences between two popular variable reduction approaches. General Psychiatry, 35(1).

M, S. and C D, J. (2021). Classification of soil fertility using machine learning-based classifier. In 2021 2nd ICSCCC, pages 138–143.

Saleh, A. et al. (2019). Theory of Ridge Regression Estimation with Applications. Wiley Series in Probability and Statistics. Wiley.

Suchithra, M. and Pai, M. L. (2020). Improving the prediction accuracy of soil nutrient classification by optimizing extreme learning machine parameters. IPA, 7(1):72–82.

Sunori, S. K. et al. (2022). Design of ann based classifiers for soil fertility of uttarakhand. In 3rd INCET, pages 1–5.

Tavares, T. R. et al. (2022). Spectral data of tropical soils using dry-chemistry techniques (vnir, xrf, and libs): A dataset for soil fertility prediction. Data in Brief, 41:108004.

Verleysen, M. and François, D. (2005). The curse of dimensionality in data mining and time series prediction. In IWANN, pages 758–770. Springer.

Wei, M. C. F. et al. (2022). Dimensionality reduction statistical models for soil attribute prediction based on raw spectral data. AI, 3(4):809–819.

Wollenhaupt, N. C. et al. (1994). Mapping soil test phosphorus and potassium for variable-rate fertilizer application. JPA, 7(4):441–448.

Yang, J. et al. (2020). Combination of convolutional neural networks and recurrent neural networks for predicting soil properties using vis–nir spectroscopy. Geoderma, 380:114616.
Publicado
08/11/2023
FIGUERÊDO, José Solenir L.; FERREIRA, Marcos Eduardo de C.; CALUMBY, Rodrigo T.. Machine Learning for Soil Attribute Prediction: An Effectiveness and Dimensionality Reduction Analysis. In: CONGRESSO BRASILEIRO DE AGROINFORMÁTICA (SBIAGRO), 14. , 2023, Natal/RN. Anais [...]. Porto Alegre: Sociedade Brasileira de Computação, 2023 . p. 302-309. ISSN 2177-9724. DOI: https://doi.org/10.5753/sbiagro.2023.26572.