Water Demand Forecasting Model Application

Abstract

Forecasting water demand requires quantifying potential relationships between relevant statistics and ambient conditions such as water price and weather. Dr Enouy (2018) demonstrates that discrete histograms can be parameterized into continuous probability density functions. Consistent parametrization allows regression analysis to be applied to the PDF statistics, thus able to reproduce PDFs through time. This work briefly introduces Dr Enouy’s (2018) methodology and mainly investigates the applicability of this method. It formalizes the implementation details of residential water application in terms of data culling, optimization and regression analysis. A modified version of this method is employed as an adaptation to the analysis of commercial water demand. This thesis also discusses the possibility of employing the scheme of software development, to assure the robustness and correctness of this implementation.