Developing groundwater potentiality models by coupling ensemble machine learning algorithms and statistical techniques for sustainable groundwater management

Abstract

The present study intends to construct a logistic regression based hybrid EML model by considering nine standalone and ensemble machine learning (EML) algorithms as parameters for groundwater potentiality mapping (GPM) in the Teesta River basin of Bangladesh, including random forest (RF), M5P, artificial neural network (ANN), random subspace (RS), dagging, bagging, random tree, support vector machine (SVM), and REPTree. The GPM were then validated using the receiver operating characteristics (ROC) curve. To investigate the effect of the parameters for GPM, we used classification and regression tree (CART) and RF based sensitivity analysis. The very high (831-1200km²) and high groundwater potential areas (521-680km²), were predicted using nine EML algorithms and one hybrid model. The REPTree (AUC-0.893) model outperformed other nine models based on ROC's area under curve (AUC). Furthermore, the LR algorithm-based hybrid EML model outperformed the REPTree model in terms of precision (AUC: 0.933).

Keywords:

• Groundwater potentiality
• statistical models
• Data mining
• hybrid ensemble model
• Sensitivity analysis

Data availability

The data that support the findings of this study are available from the corresponding author, [[email protected]], upon reasonable request.

Disclosure statement

No potential conflict of interest was reported by the authors.

Additional information

Funding

This article was funded by Deanship of Scientific Research at King Khalid University.