The Eastern Himalayas is one of the most adversely affected areas by landslides as every year there happens to be a considerable amount of human lives lost along with economic infrastructural damage. This study was performed with a quest to analyze the landslide susceptibility of Darjeeling-Kalimpong Hills, which is a crucial part of the Eastern Himalayas, through heterogeneous data-driven modeling techniques. The researchers utilized twenty geo-environmental factors, and a total of 1888 landslide points to prepare the landslide susceptibility map. The data-driven models taken into consideration were the multivariate additive regression spline (MARS), Bayesian generalized linear model (BGLM), Boosted Decision Tree (BDT), and their ensemble as MARS-BGLM-BDT which possessed good performance as measured by receiver operation characteristics (ROC) curve value of 0.86, 0.80, 0.88, and 0.95, respectively with training data. With testing data, the ROC curve values were found to be 0.89, 0.85, 0.88, and 0.94 for MARS, BGLM, BDT, and MARS-BGLM-BDT models, respectively. With the testing dataset, the ensembled MARS-BGLM-BDT model outperformed the rest of the standalone models with a promising ROC value of 0.94. The outcomes from the susceptibility maps can be utilized as a ready reference to the decision-makers at the regional level so that the unanticipated loss of human lives and economic infrastructures can be decreased. The outcome from the susceptibility analysis can be considered a prominent input to develop an early warning system for the catastrophic scenario due to landslides.