From Fuzzy-TOPSIS to Machine Learning - A Holistic Approach to Understanding Groundwater Fluoride Contamination - Abstract

Aug 5, 2026

Fluoride (F-) contamination of groundwater threatens public health worldwide and in India, with a focus on Dhanbad. The study investigates spatial distribution and contamination sources to support tailored mitigation strategies. A triad of Multi Criteria Decision Making models (Fuzzy-TOPSIS), machine learning algorithms (logistic regression, CART, and Random Forest), and classical methods is applied to 283 groundwater samples classified using WHO’s 1.5 ppm drinking-water limit.

LanguageEnglish

CategoryResearch & Report

FormatPDF

Pages54

Reading Time2 h 16 min

AccessFree to download

Abstract

Keywords

Highlights

Introduction

1 From Fuzzy-TOPSIS to Machine Learning: A Holistic Approach to Understanding 2 Groundwater Fluoride Contamination 3 Rupsha Nandi 1, Sandip Mondal2, Jajati Mandal3, Pradip Bhattacharyya 1 4 1Agricultural and Ecological Research Unit, Indian Statistical Institute, Giridih, Jharkhand, 5 India, 815301 6 2Department of Plant Pathology, The Ohio State University, Ohio, Columbus, USA, 43210 7 3 School of Sciences, Engineering & Environment, University of Salford, Manchester M5 4WT, 8 UK 9 * [Corresponding author: pradip.bhattacharyya@gmail.com](Corresponding author: pradip.bhattacharyya@gmail.com) 10 11 12 13 14 15 16 17 18 19 20 21 22 23 Graphical abstract 24 25 26 27 28 Abstract 29 Fluoride (F- ) contamination of groundwater is a prevalent environmental issue threatening 30 public health worldwide and in India. This study targets an investigation into spatial 31 distribution and contamination sources of fluoride in Dhanbad, India, to help develop tailored 32 mitigation strategies. A triad of Multi Criteria Decision Making (MCDM) models (Fuzzy- 33 TOPSIS), machine learning algorithms {logistic regression (LR), classification and regression 34 tree (CART), Random Forest (RF)}, and classical methods has been undertaken here. 35 Groundwater samples (n = 283) were collected for the purpose. Based on permissible limit (1.5 36 ppm) of fluoride in drinking water as set by the World Health Organization, samples were 37 categorized as Unsafe (n=67) and Safe (n=216) groups. Mean fluoride concentration in Safe 38 (0.63±0.02 ppm) and Unsafe (3.69 ± 0.3 ppm) groups differed significantly (t-value = -10.04, 39 p< 0.05) . Physicochemical parameters (pH, electrical conductivity, total dissolved solids, total 40 hardness, NO3-, HCO3-, SO42-, Cl-, Ca2+, Mg2+, K+, Na+ and F- ) were recorded from samples of 41 each group. The samples from ‘Unsafe group’ showed alkaline pH, the abundance of Na+ and 42 HCO3- ions, prolonged rock water interaction in the aquifer, silicate weathering, carbonate 43 dissolution, lack of Ca2+ and calcite precipitation which together facilitated the F- abundance. 44 Aspatial distribution map of F- contamination was created, pinpointing the "contaminated 45 pockets." Fuzzy-TOPSIS identified that samples from group Safe were closer to the ideal 46 solution. Among these models, the LR proved superior, achieving the highest AUC score of 47 95.6 % compared to RF (91.3 %) followed by CART (69.4 %) . This study successfully 48 identified the primary contributors to F- contamination in groundwater and the developed 49 models can help predicting fluoride contamination in other areas. The combination of different 50 methodologies ( Fuzzy-TOPSIS, machine learning algorithms, and classical methods) results 51 in a synergistic effect where the strengths of each approach compensate for the limitations of 52 the other. 53 54 Keywords: Fluoride; Hydrogeochemistry; Water quality index; Health risk assessment; 55 Logistic regression; Fuzzy-TOPSIS 56 57 58 59 60 Highlights 61 • Fluoride (F->1.5 ppm) causes fluorosis and its sources are mostly geogenic. 62 • Silicate weathering and carbonate dissolution increase groundwater levels ofF- . 63 • F- contaminated water pose non carcinogenic health risks to humans. 64 • Calcite precipitation and HCO3- abundance facilitates increase in groundwater F- . 65 • Logistic regression predicts fluoride with high accuracy (AUC=0.96) . 66 • Fuzzy TOPSIS provides evidence ofunsuitability ofF- contaminated water. 67 68 1. Introduction 69 Globally groundwater meets ~69% of agricultural, ~22% of industrial and ~8% of 70 household demands (Das et al., 2020; Jha and Tripathi, 2021) . More than 1.5 billion human 71 lives depend on groundwater solely for drinking purposes (Adimalla et al., 2019) . Drinking 72 safe water is the basic right of all living beings and serves as a key developmental indicat

FAQs

What is the main goal of the study on groundwater fluoride contamination?

To investigate the spatial distribution and likely sources of fluoride contamination in Dhanbad so that targeted mitigation strategies can be developed.

How were groundwater samples classified in the research?

Samples were categorized as Safe or Unsafe based on the WHO permissible fluoride limit of 1.5 ppm, using 283 collected groundwater samples.

Which modeling approach performed best for predicting fluoride contamination?

Logistic regression outperformed the other methods, achieving the highest AUC value (95.6%) compared with Random Forest (91.3%) and CART (69.4%).

Aug 5, 2026

Upload by Emma Wilson