Regression Analysis of Soil Compaction Parameters Using Support Vector Method

Abstract

Some challenging studies are experimentally applied for characterizing parameters in Proctor compaction tests. Compression of a fill is mechanically done in Compaction process. Compaction is a physical process which gets the soil into a dense state. Improving the shear strength and decreasing the compressibility and permeability of the soil can be done with this physical process. Support Vector Machine (SVM) is a popular method due to its performance today. This method is commonly employed in the regression analysis as well as being used in the classification process. In this study, SVM was employed to predict of compaction parameters (maximum dry unit weight and optimum moisture content) without making any experiments in a soil laboratory. In the study, more than a hundred compaction data collected from the small dams in central Anatolia region was employed. In the study, R errors are satisfied (0.92 and 0.89) for SVM models. Consequently, the proposed regression analysis with SVM is useful for model design of the projects in where there are limitations as financial and temporal.