Log logistic distribution to model water demand data

Abstract

There had been insufficient studies previously to conclude the suitability of the appropriate probability distribution functions in modelling water demand. The purpose of this study is to find an appropriate probability density function to apply in simulating water demand using real water consumption data. Daily water consumption data for four years obtained from a water company in UK and analysed using normal, log normal, log logistic and Weibull distributions and a comparison on the applicability of each distribution was assessed. Statistical modelling was performed using MINITAB. The Anderson Darling (AD) statistic was used as the goodness of fit parameter in the analysis.