Water quality index prediction using artificial neural network: a case study of Selangor River, Malaysia

Rapid urban development often leads to deterioration of river water quality, and water quality index (WQI) is a number that represents the water quality of a water body. According to Department of Environment, Malaysia parameters used to calculate WQI are dissolved oxygen (DO), biochemical oxygen demand (BOD), chemical oxygen demand (COD), potential of hydrogen (pH), suspended solids (SS) and ammoniacal nitrogen (AN). Data was collected from the ASEAN Working Group on Water Resources Management website in Sungai-Selangor section. Two ANN models were developed; a prediction model to predict the current WQI, and a forecasting model to predict the future WQI. The prediction model gave good results with very low overall root mean squared error (1.15), an excellent overall regression value (0.97874), and a high correlation with the actual WQI (99.94%). The forecasting model did not provide good result with the overall RMSE of 4.80 and overall regression value of 0.752.