This study introduces an innovative approach to slope stability analysis through the development and application of machine learning-based regression models. Analyzing a rich dataset of diverse cases, the research emphasizes non-homogeneous cohesive slopes, drawing data from the Limit Equilibrium Method (LEM). The challenge of slope failure, a global concern, necessitates a deep understanding of the underlying factors responsible for instability. Accordingly, this study identifies critical parameters, including cohesion, specific gravity, slope angle, thickness of layers, internal angle of friction, saturation condition, wind and rain, blasting conditions, and cloud burst conditions. These parameters, encompassing internal, external, and geometric aspects of the slope, are integral to this analysis. Leveraging regression algorithms such as Linear Regression, Regularized Regression, Decision Tree, Random Forest, Support Vector Regression, K-Nearest Neighbor (KNN) and XGBoost regression, the research offers robust and nuanced insights into the stability of slopes. Significantly, the study goes beyond theoretical modeling, ensuring real-life validation in actual opencast mines, and providing due importance to understanding the feature importance of each input variable. This comprehensive approach ensures not only predictive accuracy but also provides actionable insights for real-world applications in mitigating slope failure risks. The findings from this research contribute to the evolving field of machine learning in mining and present an important step forward in the pursuit of safe and sustainable mining practices.