<![CDATA[Clean and safe drinking water is a basic need for society. Based on clean water statistical data from Statistic Indonesia (BPS), the average percentage of households in West Kalimantan Province with access to safe drinking water sources in urban areas was 89.21%, while in rural areas it was 77.76% in 2023. With a population growth that outpaces the available reserves of safe drinking water, there is a clear need to monitor and analyze water quality more carefully and on a data-driven basis. This research aims to design and implement a drinking water suitability classification system using the Random Forest method as the main classification algorithm, with Principal Component Analysis (PCA) serving as the data dimensionality reduction technique. The interactive visualization was built using Python and Streamlit, taking into account User Interface (UI) and User Experience (UX) aspects, which are divided into two access levels: Public Access (for the community) and Officer Access (for agencies). Water quality data is identified based on standards from the Ministry of Health (Permenkes No. 492/MENKES/PER/IV/2010), reduced in dimensionality using PCA, and subsequently classified for its suitability using Random Forest. The test results demonstrate that this system is capable of classifying the suitability status with an accuracy of 95.35%. to ensure the model does not experience overfitting, K-Fold Cross-Validation (5-Fold) testing was conducted, yielding an average accuracy of 94.94% and proving the model’s stability. Furthermore, the User Acceptance Testing (UAT) results indicate an excellent level of user acceptance regarding the bulk upload feature and the monitoring of drinking water quality in a fast, accurate, and data-driven manner.]]>
Copyrights © 2026
