Uniaxial Compressive Strength (UCS) is a fundamental parameter in engineering projects, often serving as a primary input for various analyses. The direct determination of UCS requires laboratory sample preparation in accordance with established standards. However, in cases where sample extraction is unfeasible due to the rock type or field conditions, UCS must be estimated through indirect methods. Over the years, numerous rock properties such as porosity, density, P-wave velocity, and Schmidt hammer rebound value have been employed as predictor variables for UCS estimation. In this study, UCS was predicted using Schmidt hammer rebound (SHR) and Leeb hardness (HL), which are practical, cost-effective, and non-destructive testing methods. Various machine learning algorithms including Linear Regression, Ridge Regression, Lasso Regression, ElasticNet Regression, Random Forest, Gradient Boosting, and Support Vector Regression were applied for prediction. The correlation coefficient (R2) obtained from these models ranged between 0.75 and 1.00. Among the tested models, the Random Forest (RF) algorithm demonstrated the highest prediction accuracy, with validation metrics of RMSE = 1.93, MSE = 0.87, and R-2.
