Comparative approach of decision tree and CWQI analysis for classification of groundwater with a special reference to fluoride ion in drought-prone Boudh district of Odisha, India

Abstract

Major population in the Boudh district of Odisha, India basically depends upon groundwater for various household needs. The presence of underlying hard-rock terrain increases the scarcity, and fluoride content in the groundwater. Hydrochemical studies have been conducted to evaluate the quality of groundwater using GIS. A field reconnaissance survey has been done for assessing water table conditions. The presence of fluoride in groundwater varies from 0.239 to 4.20 mg/L before the monsoon and 0.184–3.75 mg/L after the monsoon period. Fluoride (F−) ion content is found to be more in samples of the pre-monsoon period than post-monsoon due to the dilution in the rainy season. The statistical analysis has directed that fluoride ion has a positive correlation with bicarbonate (HCO3), TA, Sodium (Na) and pH. The suitability of this groundwater is further examined and analysed with World Health Organization (WHO) standard. Various groundwater samples were cross-examined with high-end analytical techniques. A computational decision tree approach of data mining with an accuracy of 92.68% has been used to confirm the classification result of groundwater of different categories calculated through statistical analysis. The experiment through decision tree (J48) classification algorithm and Canadian water quality index (CWQI) concluded that 2 areas contain good quality of groundwater, 19 areas contain the fair quality of groundwater, 70 areas are considered marginal for drinking and only 11 areas contain poor water quality.