Monitoring the Quality Compliance of Drinking Water Treatment Process using Machine Learning Techniques

Résumé

Abstract This article aimed at measuring the extent to which chemical inputs used in the water treatment process complied with acceptable limits set by competent authorities. The computed drinking water quality index (DWQI) was visualized in a dashboard created in Microsoft Power Business Intelligence. The adopted operational approach helped compare daily compliance versus non-compliance. Then, a logistic regression predictive model was used to determine whether there were any significant separations between water treatment plants (WTPs) as per quality compliance results. The analysis was done on 13,158 daily water transactions recorded by 18 WTPs managed by the Water and Sanitation Corporation (WASAC) of Rwanda. As per DWQI values, all the WTP’s water productions generally complied with the quality requirements despite few exceptions. The predictive model accuracy was confirmed at 93.4% with all variables being statistically influential. In terms of predicted probabilities, WTPs fell in two clusters: WTPs located in the Northern, Kigali and Western provinces and those in the Eastern and Southern provinces. The first cluster had the probability of more than 95% to produce excellent water while the other had the probability of less than 95%. On average, the water treated in the provinces of the first cluster has more than 95% chance to be excellent. On the other hand, WTPs in Eastern and Southern Provinces have more than 90% but less than 95% chance to produce A-ranked water. Finally, the performance monitoring of water products was recommended to complement the current operational approach to confirm its suitability for human consumption.

Citer ce document

Accès au document

Ce document est hébergé sur son dépôt institutionnel d'origine.

Auteur(s)

Statistiques

Consultations : 1

Téléchargements : 0