Applicability of Data Mining Techniques for Predicting Electrical Resistivity of Soils Based on Thermal Resistivity

This article adopts two data mining techniques, support vector machine (SVM) and least-squares support vector machine (LSSVM), for prediction of soil electrical resistivity based on soil properties and thermal resistivity. Two models (Model I and Model II) are developed. Model I uses the percentage sum of the gravel-size and sand-size fractions (%) and thermal resistivity (RT) as input parameters. Model II uses gravel-size and sand-size fractions (%), degree of saturation (%), and thermal resistivity (RT) as input parameters. Equations have been also developed for the determination of the soil electrical resistivity (RE) of soils. The results are compared with an artificial neural network (ANN) model. This study proves the capability of SVM and LSSVM for prediction of the RE of soils.