Treatment of Surface Water Using <i>Chrysopogon zizanioides</i> With Water Quality Monitoring and Classification by Machine Learning Models

ABSTRACT This study investigates the treatment efficiency of Chrysopogon zizanioides (vetiver grass) in improving surface water quality through phytoremediation, complemented by machine learning‐based water quality classification. Water quality index (WQI) and water quality classification (WQC) were anticipated and classified using a machine learning model by filtering used Chrysopogon zizanioides . Surface water samples were collected and monitored at regular intervals every 7 days over a period of 12 weeks to ensure sufficient temporal resolution for tracking dynamic changes in key parameters such as pH, turbidity, hardness, chloride, sulfate, calcium, TDS, TSS, iron, and copper. The results show that the proposed models accurately estimate the WQI and categorize water quality with improved robustness. The study found that the NARNN model predicted WQI values better than the LSTM model. Additionally, the XGBOOST algorithm had a maximum accuracy rate of 97.01%. It also has a 99.23% sensitivity rate, confirming positive detection. With a specificity rating of 97.78%, the system accurately identified negative events. The precision rate was 94.93%, suggesting its ability to predict positive events. Finally, the algorithm's F ‐score of 98.54% indicates its WQC prediction performance. This study integrates ecological restoration, technology, and community‐focused solutions to provide clean water access and sustainable water management to achieve the United Nations Sustainable Development Goals (UN SDGs 6).