Reliable water quality prediction and parametric analysis using explainable AI models

Area of Science:

• Environmental Science
• Computer Science
• Public Health

Background:

• Water quality is crucial for public health and environmental safety.
• Contaminated water poses significant health risks.
• Monitoring and managing water purity, especially Total Dissolved Solids (TDS), is essential.

Purpose of the Study:

• To automate water quality estimation using Artificial Intelligence (AI).
• To employ Explainable AI (XAI) for identifying significant water quality parameters.
• To enhance transparency in machine learning-based water quality classification.

Main Methods:

• Utilized various machine learning (ML) models including Logistic Regression, Support Vector Machine (SVM), Gaussian Naive Bayes, Decision Tree (DT), and Random Forest (RF).
• Applied Explainable AI (XAI) techniques (SHAPELY explainer) for feature importance and prediction justification.
• Evaluated model performance using metrics like Accuracy, F1-Score, Precision, and Recall.

Main Results:

• Random Forest (RF) classifier achieved optimal performance.
• Achieved an Accuracy of 0.9999, F1-Score of 0.9999, Precision of 0.9997, and Recall of 0.998.
• XAI methods provided clear explanations for water quality estimations.

Conclusions:

• The study successfully demonstrated automated water quality estimation with high accuracy.
• XAI enhances the interpretability of ML models in water quality assessment.
• This research offers a vision for future water quality management and safety.