Synthetic Dataset for Non-Linear Rock Mass Rating (RMR) Prediction using Machine Learning

This dataset contains 2000 samples of synthetic rock mass parameters generated via Monte Carlo simulation to test machine learning models for Rock Mass Rating (RMR) prediction. The dataset was designed to simulate non-linear interactions found in geological environments, specifically the combined negative effect of groundwater and poor discontinuity conditions, which traditional linear RMR summation methods fail to capture. The repository includes the generated dataset (.csv) and the Python script (.py) used to generate it, ensuring full transparency and reproducibility of the methodology. Variables (Columns): Input Parameters (5 Basic RMR Parameters): UCS_MPa: Uniaxial Compressive Strength (5-250 MPa). RQD_Percent: Rock Quality Designation (10-100%). Spacing_mm: Discontinuity Spacing (20-4000 mm). Condition_Score: Discontinuity Condition based on Bieniawski (1989) (0-30 points). Groundwater_Score: Groundwater conditions (0, 4, 7, 10, 15 points). Output / Target Variables: Classic_RMR: The theoretical RMR value calculated by simple arithmetic summation of the five inputs. Actual_RMR_Field: The simulated "Ground Truth" value, incorporating non-linear penalty functions for water-clay interactions and Gaussian noise (mean=0, std=5) to represent field uncertainties. This dataset supports the study: "Modeling Non-Linear Relationships in Rock Mass Rating (RMR) System: A Methodological Stacking-Based Hybrid Approach".