Explainable Ensemble Machine Learning Framework for Multi-Class Ecotoxicity and Drinkability Prediction of Groundwater Using Hydro-Chemical Indicators

作者
S Vengatesh Kumar,Onkar Nath Thakur,Nitesh Singh Bhati
出处
期刊:Research Square
标识
DOI:10.21203/rs.3.rs-9315913/v1
摘要

Abstract Public health and ecological systems and safe drinking water supplies face serious dangers from groundwater contamination which particularly affects areas that experience rapid urban growth and industrial development. Groundwater quality assessment methods which scientists have traditionally used depend on laboratory testing and statistical analysis which takes extensive time while generating high costs and fails to produce accurate predictive results. This study introduces HydroNet-X, an innovative explainable ensemble deep learning framework which provides both precise results and understandable insights for three different groundwater ecotoxicity and drinking water safety assessments based on hydro-chemical data. The HydroNet-X system uses SHAP-based explainable artificial intelligence to improve prediction results and model understanding through its integration of Extreme Gradient Boosting and Adaptive Boosting and Polynomial Support Vector Machine and optimized ensemble learning. The model used 13 hydro-chemical parameters which meet WHO water quality standards to train and test its performance on a real-world dataset that included 1,345 groundwater samples from Madhya Pradesh India. The Water Quality Index (WQI) assessment produced four different drinkability classifications which included Excellent, Good, Poor, and Bad. The HydroNet-X model achieved exceptional classification accuracy with 99.2% which exceeded the performance of existing models such as XGBoost (98%) and Support Vector Machine (97%) and Decision Tree (96%) and Logistic Regression (93%) and AdaBoost (76%). The proposed framework demonstrated strong predictive performance through its achievement of an average F1-score of 0.994 and precision of 0.996 and recall of 0.995. The SHAP explainability analysis showed that Electrical Conductivity (EC) and Total Dissolved Solids (TDS) and Total Hardness (TH) serve as primary hydro-chemical indicators which determine groundwater ecotoxicity and drinkability assessment. The HydroNet-X framework delivers an accurate solution which can expand its capacity while maintaining its ability to monitor groundwater quality and assess environmental risks. The system enables sustainable water resource management through its decision support capabilities.

科研通智能强力驱动
Strongly Powered by AbleSci AI
科研通是完全免费的文献互助平台,具备全网最快的应助速度,最高的求助完成率。 对每一个文献求助,科研通都将尽心尽力,给求助人一个满意的交代。
实时播报
刚刚
刚刚
wangxiangqin发布了新的文献求助10
1秒前
luo发布了新的文献求助10
1秒前
ws完成签到,获得积分10
1秒前
FossilGitter8完成签到,获得积分10
1秒前
灵巧谷波完成签到,获得积分10
2秒前
信仰完成签到,获得积分10
2秒前
英俊的铭的应助被冉123采纳,获得10
4秒前
4秒前
科研通AI65的应助被白糖采纳,获得10
4秒前
xiaoxun369完成签到 ,获得积分20
6秒前
小马甲的应助被榴莲采纳,获得30
6秒前
无限夏之完成签到,获得积分10
7秒前
信仰发布了新的文献求助10
7秒前
秦艽发布了新的文献求助10
7秒前
莓啤汽完成签到 ,获得积分10
8秒前
9秒前
隐形曼青的应助被cc采纳,获得30
9秒前
悠然地八音完成签到,获得积分10
9秒前
自觉的绮烟完成签到,获得积分10
10秒前
LiuShenglan完成签到,获得积分10
10秒前
10秒前
送你一朵小红花完成签到,获得积分20
10秒前
11秒前
想过完成签到 ,获得积分10
11秒前
山河发布了新的文献求助10
11秒前
wangxiangqin完成签到,获得积分10
11秒前
秋风的应助被负责的问雁采纳,获得10
12秒前
小y完成签到,获得积分10
12秒前
12秒前
13秒前
宝贝是小咸鱼完成签到,获得积分10
13秒前
冷宇完成签到 ,获得积分10
14秒前
平p完成签到,获得积分10
14秒前
Cherish完成签到,获得积分20
14秒前
yanyan发布了新的文献求助10
14秒前
Akim的应助被qiuu采纳,获得10
14秒前
林读书发布了新的文献求助10
16秒前
16秒前
高分求助中
(应助此贴封号)【重要!!请各用户(尤其是新用户)详细阅读】【科研通的精品贴汇总】 10000
拉伸高分子熔体中缠结行为的模拟和小角中子散射研究 1700
The World Atlas of Wine 9th edition 1000
奈特基础组织学:配套组织病理学(第4版)(Netter's Essential Histology With Correlated Histopathology) 800
八旗制度の研究 600
Boundaries and Bodhisattvas 500
高分子基功能界面水蒸发材料与应用研究 500
热门求助领域 (近24小时)
化学 材料科学 医学 生物 纳米技术 计算机科学 工程类 有机化学 化学工程 生物化学 物理 复合材料 内科学 催化作用 光电子学 人工智能 细胞生物学 无机化学 物理化学 数学
热门帖子
关注 科研通微信公众号,转发送积分 8182014
求助须知:如何正确求助?哪些是违规求助? 9571728
关于积分的说明 21819478
捐赠科研通 7778441
什么是DOI,文献DOI怎么找? 3433580
关于科研通互助平台的介绍 2557781
邀请新用户注册赠送积分活动 2479640