Machine Learning Approach for Evaluating water Safety for human consumption in Dili city, Timor-Leste

Drinking water is essential for human survival and the sustainability of all life on Earth. In developing countries like Timor-Leste, government-provided water services are severely limited and face numerous challenges, including low service standards, poor water quality, and insufficient funding. In Timor-Leste, 80% of the population is dependent on agriculture. In Dili, 70% of the water supply comes from groundwater and 30% from surface water. As the population grows, the demand for water increases, affecting both its availability and quality. Efforts are underway to address these issues and improve the safety of water for drinking and other domestic uses. In Dili, 46 boreholes are used for water supply, has been used in this study, but ten are over-exploited and fifteen are unfit due to microbiological contamination, manganese, iron, TDS and hardness of water. Machine learning (ML) has shown promise in solving real-world challenges, including water quality assessment. In this paper, three ML algorithms-K-Nearest Neighbors (KNN), Decision Trees, and Naive Bayes (NB)-are applied to assess the potability of water. Performance evaluation and k-fold cross-validation were used to assess the effectiveness of the models. The KNN algorithm showed the highest accuracy, reaching 97%. Thus, in the future water quality prediction in Timor-Leste is necessary to incorporate environmental data and sewage systems.