2025

European Research Studies Journal

Rocznik: 2025 | Tom: 28 | Numer: 3

artykuł naukowy

angielski

EN The purpose of this article is to develop a framework for reducing water quality assessment parameters while maintaining the precision and accuracy of evaluation. Design/Methodology/Approach: The study was based on machine learning methods, including classification and feature selection techniques. Specifically, data discretization, correlation-based feature selection, and the following classification algorithms were applied: artificial neural network, decision tree, random forest, and support vector machine. The study also utilized two datasets describing water quality based on physical, chemical, and biological parameters. Findings: The conducted research indicates that the correlation-based feature selection filter, combined with other machine learning methods, is an effective tool that enables a substantial reduction in the number of parameters used for water quality assessment without any loss of accuracy or precision. Practical Implications: The study’s findings make it possible to optimize the water quality assessment process by reducing the number of required chemical, physical, and biological tests and analyses. This reduction can lower the costs of data collection, analysis, and interpretation, minimize data gaps, and increase monitoring frequency – thereby economically rationalizing water quality testing procedures. Originality/Value: The development of a framework for reducing water quality assessment parameters while maintaining precision and accuracy constitutes a contribution to the field of water quality assessment methods and the economic rationalization of water quality testing.

1538 - 1552

10.35808/ersj/4248

https://ersj.eu/journal/4248

CC BY (uznanie autorstwa)

otwarte czasopismo

ostateczna wersja opublikowana

12.2025

w momencie opublikowania

100