The Western Himalayas in India have experienced increased debris flow hazards due to intensified precipitation and landslides. This study investigates the dynamics of debris flow deposition using an experimental flume setup, simulating conditions in the Western Indian Himalayas. A total of 32 experiments were conducted, manipulating water content and the proportion of stony particles in debris mixtures. The study utilized machine learning techniques, particularly XGBoost, to analyze the feature importance of various parameters on flow characteristics, including deposit length, width, and thickness. The findings highlighted the significant influence of stony particles, with a composition of 8–12% markedly impacting deposit thickness and width. Additionally, water content played a critical role, negatively affecting deposit thickness while enhancing mobility. The high predictive accuracy of the XGBoost model underscores the importance of understanding debris flow mechanisms to mitigate geohazard risks. This research contributes valuable insights into the morphodynamic processes governing debris flow deposition, providing a foundation for more effective disaster mitigation strategies in the region.