The reliability of data-driven approaches in generating landslide susceptibility maps depends on data quality, analytical method selection, and sampling techniques. Selecting optimal datasets and determining the most effective analytical methods pose significant challenges. This study assesses the performance of seven machine learning classifiers in the Himalayan region of the China–Pakistan Economic Corridor, utilizing statistical techniques and validation metrics. Thirteen geo-environmental variables were analyzed, including topographic (8), land cover (1), hydrological (1), geological (2), and meteorological (1) factors. These variables were evaluated for multicollinearity, feature importance, and their influence on landslide incidences. Our findings indicate that Support Vector Machines and Logistic Regression were highly effective, particularly near fault zones and roads, due to their effectiveness in handling complex, non-linear terrain interactions. Conversely, Random Forest and Logistic Regression demonstrated variability in their results. Each model distinctly identified landslide susceptibility zones ranging from very low to very high risk. Significant conditioning variables such as elevation, rainfall, lithology, slope, and land use were identified, reflecting the unique geomorphological conditions of the Himalayas. Further analysis using the Variance Inflation Factor and Pearson correlation coefficient showed minimal multicollinearity among the variables. Moreover, evaluations of Area Under the Receiver Operating Characteristic Curve (AUC-ROC) values confirmed the strong predictive capabilities of the models, with the Random Forest Classifier performing exceptionally well, achieving an AUC of 0.96 and an F-Score of 0.86. This study shows the importance of model selection based on dataset characteristics to enhance decision-making and strategy effectiveness.

The effectiveness of data-driven landslide susceptibility mapping relies on data integrity and advanced geospatial analysis; however, selecting the most suitable method and identifying key regional factors remains a challenging task. To address this, this study assessed the performance of six machine learning models, including Convolutional Neural Networks (CNNs), Random Forest (RF), Categorical Boosting (CatBoost), their CNN-based hybrid models (CNN+RF and CNN+CatBoost), and a Stacking Ensemble (SE) combining CNN, RF, and CatBoost in mapping landslide susceptibility along the Karakoram Highway in northern Pakistan. Twelve geospatial factors were examined, categorized into Topography/Geomorphology, Land Cover/Vegetation, Geology, Hydrology, and Anthropogenic Influence. A detailed landslide inventory of 272 occurrences was compiled to train the models. The proposed stacking ensemble and hybrid models improve landslide susceptibility modeling, with the stacking ensemble achieving an AUC of 0.91. Hybrid modeling enhances accuracy, with CNN–RF boosting RF’s AUC from 0.85 to 0.89 and CNN–CatBoost increasing CatBoost’s AUC from 0.87 to 0.90. Chi-square (χ²) values (9.8–21.2) and p-values (<0.005) confirm statistical significance across models. This study identifies approximately 20.70% of the area as from high to very high risk, with the SE model excelling in detecting high-risk zones. Key factors influencing landslide susceptibility showed slight variations across the models, while multicollinearity among variables remained minimal. The proposed modeling approach reduces uncertainties, enhances prediction accuracy, and supports decision-makers in implementing effective landslide mitigation strategies.