| Summary: | Water crises are often experienced by many developing countries worldwide. Predicting future domestic water demand and identifying the influential factors are vital to managing water supply effectively. This study aims to determine the best predictive models among Multiple Linear Regression (MLR), Multi-layer Perceptron (MLP), and Radial Basis Function (RBF) Neural Networks as well as to identify the significant influential factors towards domestic water demand. Based on the yearly records from 2000 to 2018 obtained from the Malaysian Water Association, the Department of Environment, and the Department of Statistics Malaysia the analysis results indicate an increasing pattern of domestic water in Malaysia with the demand for non-domestic water twice lower than domestic water. Based on RMSE and R-squared, Multi-layer Perceptron is the best model for predicting domestic water demand. The MLR model shows that the two most significant influential factors towards domestic water demand are price and design capacity, with negative and positive relationships. The results describe that an increase in price affects a decrease in water demand, while an increase in design capacity will reduce the water demand. The findings suggest that the water utilities in Malaysia should focus more on these identified factors. © (2024) UMT Press
|
|---|