Comparative analysis of support vector machine and artificial neural network models for soil cation exchange capacity prediction

The aim of this study was to compare the performance of support vector machine and artificial neural network techniques to predict the soil cation exchange capacity of an agricultural research station in terms of soil characteristics (clay, silt, sand, gypsum, organic matter). The data consist of 380 soil samples collected from different horizons of 80 soil profiles located in the Khoja (Khajeh) region of Azerbaijani provinces, Iran. The support vector machine and artificial neural network models predict the cation exchange capacity from the above soil characteristics of the samples. The models' results are compared using three criteria, i.e., root-mean-square errors, Nash-Sutcliffe and the correlation coefficient. A comparison of support vector machine results with artificial neural network method indicates that artificial neural network is better than the support vector machine method in prediction of the cation exchange capacity.