ESTIMATION OF UNIAXIAL COMPRESSIVE STRENGTH BASED ON REGRESSION TREE MODELS

Abstract

This paper presents the estimation of the uniaxial compressive strength for mudstone and wackestone carbonates. The need for the estimation has occurred due to inability to fulfill the high quality requirements of sample treatment during direct determination of this physical and mechanical property on certain types of rocks. For the needs of modelling intact rock materials, extracted from six locations in Croatia, were tested. The following properties were examined: density, effective porosity, point load strength index, Schmidt rebound hardness, P-wave velocity and uniaxial compressive strength which was the target value of the used statistical models. The statistical models based on multiple linear regression and regression trees were considered and compared using cross validation, which showed that the most efficient estimation of the uniaxial compressive strength is obtained using random forests.

Keywords

• estimation
• regression tree
• random forest
• carbonates
• uniaxial compressive strength
• point load strength index
• Schmidt rebound hardness
• P-wave velocity