Comparison of multiple linear regressions and artificial intelligence-based modeling techniques for prediction the soil cation exchange capacity of Aridisols and Entisols in a semi- arid region

Abstract The cation exchange capacity (CEC) of the soil is a basic chemical property, as it has been approved that the spatial distribution of CEC is important for decisions concerning pollution prevention and crop management. Since laboratory procedures for estimating CEC are cumbersome and time-consuming, it is essential to develop an indirect approach such as pedo-transfer functions (PTFs) for prediction this parameter from more readily available soil data. The aim of this study was to compare multiple linear regressions (MLR), adaptive neuro-fuzzy inference system (ANFIS) and artificial neural network (ANN) including multi-layer perceptron (MLP) and radial basis function (RBF) models to develop PTFs for predicting CEC of Aridisols and Entisols in Khuzestan province, southwest Iran. Five soil parameters including bulk density, calcium carbonate, organic carbon, clay and silt were considered as input variables for proposed models. The prediction capability of the models was evaluated by means of comparison with observed data through various descriptive statistical indicators including mean square error (MSE