CMStatistics 2018: Start Registration
View Submission - CMStatistics
B1437
Title: Evaluating water meters performance: An industry case study Authors:  Ana Borges - Porto Polytechnic (Portugal) [presenting]
Clara Cordeiro - CEAUL and FCT, UALg (Portugal)
Regina Casimiro - Infraquinta (Portugal)
Abstract: Infraquinta is the water utility that manages the water and the wastewater services of a well-known tourist place in Algarve, Portugal. Infraquinta seeks to improve mechanisms for predictive planning based on data analysis. In particular on evaluating water meters performance by using historical data. Hence, the main purpose of this analysis is to find out the water meter performance breakpoint and when to replace it. Firstly, the decomposition of monthly time series into seasonality, trend and irregular components is performed. The nonparametric Seasonal-Trend decomposition by Loess can identify a seasonal component that changes over time, a non-linear trend and it can be robust in the presence of outliers. Secondly, the detection of structural breaks in the trend component is performed since it is the one related to the meter performance. This approach tests for structural changes in linear regression models, estimating the number of segments and the breakpoints by minimizing the Bayesian information criterion and the residual sum of squares. Outcomes shown that this methodology can detect the water meter breaking points and that if incorporated in a system that automatically analyses the data and updates the information, could be the solution sought by Infraquinta.