Modeling water consumption in public schools in Joinville, SC

Abstract

This work aims to model per capita water consumption in schools in the city of Joinville-SC. As the school environment is largely responsible for the transmission of knowledge, including information on water conservation, 26 public schools were selected for this research. A survey was carried out on the use of water, through visits and a questionnaire to the management team of each school. The data collected allowed the characterization of schools and a preliminary diagnosis of water use. A descriptive analysis of per capita water consumption was performed, with values ranging from 5.15 to 18.59 liters/student/day. Correlation analysis, multiple linear regression, Bayesian linear regression and geographically weighted regression were used. The results of the correlation analysis reveal that the higher the average income in the neighborhood where the school is located and the area per student ratio, the higher the per capita water consumption. On the other hand, the number of students has a negative correlation with per capita consumption. In the statistical modeling, the independent variable area per student ratio was significant.

Keywords:
Water consumption; Schools; Multiple linear regression; Bayesian linear regression; Geographically weighted regression.