Comparison of linear regression, Fuzzy and Fuzzy-genetic models to predict soil cation exchange capacity

Cation exchange capacity (CEC) is one of the most important soil chemical properties that affects other chemical, physical, and biological soil properties and fertility. In this study, performance of some procedures such as regression, Fuzzy and Fuzzy-genetic approaches in estimation of soil CEC has been investigated. Consequently, the required data of 770 samples from the Europe database (IES) was extracted. Then multiple linear regression, Fuzzy and Fuzzy-genetic approaches were used for development of pedotransfer functions for estimating of soil CEC. The coefficient of determination (R2), root mean square error (RMSE) and mean absolute error (MAE) criteria were used for evaluation of the proposed models. The values of R2, RMSE and MAE obtained for the linear regression model 0.72, 7.42 cmolc kg-1 and 9.13 cmolc kg-1, for Fuzzy model, 0.78, 5.44 cmolc kg-1 and 4.32 cmolc kg-1, for Fuzzy-genetic model 0.84, 4.7 cmolc kg-1 and 3.57 cmolc kg-1, respectively. These results indicated that the Fuzzy-genetic CEC model is more accurate than Fuzzy model, and Fuzzy model is more accurate than regression CEC model.