Authors:
K. B. Susha, H. Jayamangala

Addresses:
Department of Computer Applications, Vels Institute of Science, Technology and Advanced Studies, Chennai, Tamil Nadu, India.

Abstract:

Monitoring water quality is a key element to maintaining the sustainability of our environment and public health. Water-quality safety classification systems based on machine learning have been developed as effective tools for predicting potable and non-potable conditions. Nevertheless, classification performance depends significantly on the effectiveness of data preprocessing techniques before model training. Although machine learning has been increasingly used to assess water quality, few studies have systematically tested the effects of various preprocessing methods on classification results. This study compares and contrasts the most significant preprocessing methods, including missing data imputation, outlier detection, normalisation, feature selection, dimensionality reduction, and class balancing. A systematic experimental framework is created from publicly available water-quality datasets. The following classification models are evaluated: Random Forest, Support Vector Machine, XG-Boost, and Artificial Neural Networks, with varying preprocessing conditions. The experimental results show that the accuracy, precision, recall, F1-score, and computational efficiency of classification can be significantly influenced by preprocessing. Comprehensive feature selection, along with powerful normalisation and advanced imputation strategies, delivers the highest overall performance. The results offer useful recommendations for selecting appropriate pre-processing pipelines for water-quality safety classification systems and for developing reliable intelligent water-monitoring systems.

Keywords: Water Quality Classification; Data Preprocessing; Machine Learning; Feature Selection; Data Imputation; Water Potability Prediction; Environmental Monitoring.

Received on: 20/03/2025, Revised on: 27/05/2025, Accepted on: 14/08/2025, Published on: 05/06/2026

DOI: 10.69888/FTSESS.2026.000714

FMDB Transactions on Sustainable Environmental Sciences, 2026 Vol. 3 No. 2, Pages: 54-70

• Views : 19
• Downloads : 6

Download PDF