The Himalayan mountains are prone to landslide disasters, which cause injury and fatalities among people. Remote sensing, particularly interferometric synthetic aperture radar (InSAR) based analyses, may help find the surface subsidence velocities, the rate of vertical movement of the Earth’s surface downward. These subsidence velocities may help identify areas prone to landslides. Literature suggests that geotechnical parameters may contribute to understanding surface subsidence velocities. However, there is less research on developing data-driven algorithms that relate the geotechnical parameters with the subsidence velocities in an area. In this research, data-driven algorithms, relying upon geotechnical parameters measured across diverse locations in the Himalayan mountains, predict the subsidence velocities measured from InSAR analysis of open-source Sentinel-1 data of the same area. An InSAR-based displacement map of the study area is generated using the Small Baseline Subset (SBAS) algorithm. The ranking of geotechnical parameters is first conducted using several feature selection methods. Out of 23 parameters, 11 top-ranked features are selected for developing data-driven algorithms, including multiple regression, random forest, instance-based learner, an optimized version of support vector regression named sequential minimal optimization regression (SMOreg), multilayer perceptron (MLP), and an ensemble of MLP and SMOreg (MLP-SMO). These algorithms used the top-ranked geotechnical parameters and predicted the subsidence velocities. Results suggested that the MLP-SMO algorithm provided the best fit for data in 10-fold cross-validation with 0.94 RMSE. The MLP was the second-best model with 1.6 RMSE. Implications for developing subsidence velocity models using geotechnical parameters via data-driven approaches are discussed.