Enhancement of water quality index prediction using support vector machine with sensitivity analysis

For more than 25 years, the Department of Environment (DOE) Malaysia has implemented a water quality index (WQI) that uses six key parameters: dissolved oxygen (DO), biochemical demand (BOD), chemical (COD), pH, ammoniacal nitrogen (AN), and suspended solids (SS). Water analysis is an essential component resources management must be properly managed to prevent ecological damage from pollution ensure compliance with environmental regulations. This increases need define efficient method for WQI analysis. One major challenges current calculation it requires series sub-index calculations are time consuming, complex, prone error. In addition, cannot calculated if one or parameters missing. this study, optimization was developed address complexity process. The potential data-driven modeling, i.e., Support Vector Machine (SVM) based on Nu-Radial basis function 10-fold cross-validation, explored improve prediction in Langat watershed. A thorough sensitivity under scenarios also conducted determine efficiency model prediction. first scenario, SVM-WQI showed exceptional ability replicate DOE-WQI obtained statistical results at very high level (correlation coefficient, r > 0.95, Nash Sutcliffe efficiency, NSE >0.88, Willmott’s agreement, WI 0.96). second modeling process can estimated without any parameters. It seen parameter DO most important factor determining WQI. pH affects least. Moreover, three show terms cost by minimizing number variables input combination ( 0.6, >0.5 (good), 0.7 (very good)). summary, will greatly accelerate decision making data accessible attractive human intervention.