<p>Forests, often referred to as the lungs of the Earth, are increasingly suffering from multiple fire events in a warming world, heightening the risk of species loss and ecosystem disturbance. Forest fires significantly challenge the diverse life forms and ecosystem services these environments provide. This study evaluates and predicts forest fire susceptibility in the Sikkim Himalayan forests using machine learning and deep learning (Rprop + algorithm) frameworks, including a bagging-based ensemble Random Forest (RF) and a boosting-based ensemble Gradient Boosting Model, integrating the environmental variables of the forest landscape to predict fire susceptibility. Notably, the RF model outperformed the deep learning model, which required extensive datasets or variables to discern the underlying patterns. Out of the three models, RF has shown an overall accuracy of 89.43%, precision of 88.56%, recall of 90.23%, F1 Score of 89.40%, AUC of 0.894, and kappa of 0.788. Factors such as proximity to roads and wind velocity, along with anthropogenic activities such as biomass burning, land clearing, and slash-and-burn agriculture by villagers, significantly influence forest fires. Identifying high-risk zones, particularly in the western and southern districts, has critical implications for forest fire management in Sikkim. The findings of this study offer valuable insights for decision-makers, aiding in the development of appropriate mitigation measures and policies. This research underscores the importance of preserving these crucial ecosystems and establishes a foundation for proactive fire prevention strategies and sustainable ecosystem management practices.</p>