An Empirical Study of Regression Algorithms for Soil Organic Matter Prediction

Abstract

Soil organic matter (SOM) is an important component that exists in soils because it is closely related to soil health and fertility. Hence, knowing the existence of SOM in soils is crucial for management corrections. So far laboratory analysis is required for SOM determination. However, such procedures are costly and labor-time consuming. Alternative methodologies for SOM determination are needed to achieve sustainability. The rise of artificial intelligence and machine learning provide promising approaches that can be exploited for this purpose. The aim of this study is to identify the best regression algorithm for SOM prediction for citrus planted soils. Several machine learning approaches are investigated, including adaptive boosting, gradient boosting, random forest, and multi-layer perceptron neural network.