<p>Landslide Susceptibility Map (LSM) is one of the essential tools for town planner’s/ government institutions to make the decision about land use planning. Utilisation of LSM map, will prevent the anthropogenic activities to be executed at essential level at high or very high landslide susceptible locations and thus mitigate the risk due to such natural hazards. The present study integrates the two strategies for LSM modelling in which the first one is to evaluate the effect of various feature engineering techniques applied on causative factors measured at ratio scale while the another one is hyper-parameters tuning of implemented Machine Learning (ML) algorithm i.e. decision tree. All the models were created using statistical/ML techniques based on fourteen pre-processed causative factors which are acquired or derived primarily using remote sensing. These factors includes topographic, anthropogenic, and environmental features while geological factor is acquired from the open domain. The Area under Receiver Operating Characteristics and overall accuracy measures were used for assessing the performance of various implemented approaches for LSM. The study area considered for the present study is Rishikesh to Gangotri axis situated in the Uttarakhand province of India. Comparative study of such statistical and ML models using different types of feature engineering techniques as well as hyper-parameters tuning is the first time attempted and can be used in futuristic analysis of LSM generation for other geographic areas too.</p>