Scholarlink

Monitoring water quality is important to ensure the safety of potable water and preserve a healthy ecosystem. This research introduces a sophisticated machine learning framework designed to predict the Water Quality Index (WQI) of drinking water sourced from tubewells throughout West Bengal. This study analyzes water quality parameter values from different wells and dug wells of 20 districts and 835 blocks of West Bengal through machine learning to know the quality of groundwater. Analysis is done on Artificial Neural Networks, Random Forest, and XGBoost algorithms for WQI prediction based on historical water quality data. The performance of the model is measured using Mean Squared Error, where the XGBoost model performs better in comparison to Random Forest. Additionally, the ANN-based model was tested and optimized in an attempt to improve accuracy further, which resulted in higher predictive accuracy. Finally, the advanced configuration of XGB was used, that being XGBRegressor with optimal hyperparameters, which has only slightly different MSE values. Results were validated by comparing predicted WQI values with the Weighted Average Water Quality Index (WAWQI), which has shown more accurate results in accordance with the statistical model's predictions. This predictive modeling approach offers a robust solution for proactive water quality management in West Bengal.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Machine Learning-Based Approach for Prediction of Water Quality Index for Groundwater Across West Bengal

  • 会议论文

  • 发表日期：

    2026年

查看全文
  • Tirtha Chatterjee,
  • Subhankar Dey,
  • Ashimanta Sengupta,
  • Abhishek Hazra,
  • Arya Banerjee

摘要

Monitoring water quality is important to ensure the safety of potable water and preserve a healthy ecosystem. This research introduces a sophisticated machine learning framework designed to predict the Water Quality Index (WQI) of drinking water sourced from tubewells throughout West Bengal. This study analyzes water quality parameter values from different wells and dug wells of 20 districts and 835 blocks of West Bengal through machine learning to know the quality of groundwater. Analysis is done on Artificial Neural Networks, Random Forest, and XGBoost algorithms for WQI prediction based on historical water quality data. The performance of the model is measured using Mean Squared Error, where the XGBoost model performs better in comparison to Random Forest. Additionally, the ANN-based model was tested and optimized in an attempt to improve accuracy further, which resulted in higher predictive accuracy. Finally, the advanced configuration of XGB was used, that being XGBRegressor with optimal hyperparameters, which has only slightly different MSE values. Results were validated by comparing predicted WQI values with the Weighted Average Water Quality Index (WAWQI), which has shown more accurate results in accordance with the statistical model's predictions. This predictive modeling approach offers a robust solution for proactive water quality management in West Bengal.