Predictive Models for Short-Term Urban Water Demand Forecasting

The precise forecasting of urban water demand is crucial for effectively administrating water resources. This paper presents a novel approach utilizing wavelet-boosting support vector regression (SVR) to address the complex and dynamic characteristics of urban water consumption. The methodology aims to effectively capture both short-term fluctuations and seasonal patterns in water usage by leveraging the strengths of wavelets, boosting techniques, and SVR. Wavelet transformations are applied to decompose time series data into different frequency components to extract essential features. Boosting enhances the model's accuracy through the iterative training of SVR models on weighted subsets of the dataset. The effectiveness of the proposed model is evaluated using historical urban water demand data, taking into account elements such as population growth, economic trends, and climatic conditions. The performance of the model is measured for lead times of three and five days ahead, using metrics including the coefficient of determination (R<sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sup>), and root mean square error (RMSE). Research contributes to developing advanced tools for sustainable water management in urban areas, enabling efficient allocation and distribution of water resources.