ABSTRACT
Alike several other countries, Bangladesh has large dependency on groundwater for drinking, agriculture and industrial usages. This study aimed at developing a new methodology based on supervised machine learning to predict Groundwater Quality Index (GWQI) which helps to know about Groundwater Quality Class (GWQC) in groundwater samples during pre-monsoon and post-monsoon seasons. The proposed methodology contains five major inputs i.e. well depth (WD), arsenic (As), manganese (Mn), iron (Fe) and calcium (Ca) to determine GWQI and GWQC. We successfully implemented the popular supervised ML models and concluded that the role of MLR is best in both pre-monsoon and post-monsoon to predict GWQI with MAE closes to 0.03 and 0.02, respectively. We also applied various classifiers where Naïve Bayes plays a significant role with 85% accuracy in pre-monsoon and 94% accuracy in post-monsoon respectively.
