Development of a model based on Support Vector Machines to predict the degradation of pesticides in biobeds systems

Abstract

Pesticides are chemical compounds used to mitigate, reduce, or eliminate the impact of pests on agricultural production. Due to their nature, pesticides are potentially toxic to many organisms, including humans. Among the various methods used to decontaminate pesticides in soils, the use of biological beds (biobeds) is a feasible option to minimize their contamination. The main problematic to use biobeds is the difficult to predict their behavior due biotic and abiotic factors. This study focuses on the use of the support vector machine (SVM), for the generation of predictive models of pesticide degradation in biobeds systems. The results show that the Gaussian and polynomial kernel has the best performance to model experimental data. The statistical parameters of R-Squared were 0.93 for Gaussian kernel and polynomial, 0.83 for cubic, 0.76 for quadratic and 0.52 for lineal. The Gaussian model could be used to provide the characteristics to improve of pesticide degradation.