Assessing the nature of potential groundwater zones through machine learning (ML) algorithm in tropical plateau region, West Bengal, India.

Groundwater recharge is essential for managing surface and subsoil water resources. Not only for supplying people with daily drinking water, groundwater use for agricultural land and people's livelihood is also continuously increasing. As a result, there has been a decline in groundwater supply in various parts of the world, and it is highly desirable to identify the potential groundwater zones for specific sustainable development. This study aims to use an easy-to-use tool package named Landslide Sustainability Mapping Tool pack (LSM tool Pack) for preparing potential groundwater zone based on R and ArcGIS software integration. This tool uses five modules for processing. Among them, the Feature selection (FS) module brings a novel approach, determining the best subset feature for demarcating the groundwater potential zone. As a result, this best factor subset is used as an input of this tool pack. Additionally, PE modules evaluate the performance of proposed models in statistical performance metrics. In addition, the receiver operating characteristic (ROC) curve was obtained with the integration of Performance Evaluation (PE) modules and ARC maps, which helps visual interpretation in evaluating models. This study uses the LSM tool Pack in the Rupnarayan river basin to map the potential groundwater zone based on fourteen controlling factors selected through the FS module, which will further help the local government to make a substitute policy. [ABSTRACT FROM AUTHOR]