ABSTRACT

Urban water consumption can be analyzed by means of applied research to help promote the development of environmentally sustainable cities. In this study, water consumption models are established for multifamily residential buildings in Southern Brazil using multiple linear regression analysis and cross-validation techniques. Sociodemographic and building characteristics were analyzed in a sample of 78 buildings for a period of 25 months. Six forecast models were obtained which met the assumptions of constant error variance, multicollinearity, autocorrelation, residual diagnosis, and generalization ability, explaining the variations in up to 90.61% of the building monthly water consumption and 68.72% of the per capita daily water consumption. The most significant variables were building age, number of stories, total area per floor and per person, dwellers per room and per apartment, number of bedrooms, type of metering system, presence of swimming pools and presence of alternative water-supply systems.

摘要

可以通过应用研究来分析城市用水量，以帮助促进环境可持续城市的发展。在这项研究中，使用多元线性回归分析和交叉验证技术为巴西南部的多户住宅建筑建立了耗水量模型。在为期 25 个月的时间里，对 78 座建筑的样本进行了社会人口学和建筑特征分析。得到了满足恒定误差方差、多重共线性、自相关、残差诊断和泛化能力假设的6个预测模型，解释了高达90.61%的建筑月用水量和68.72%的人均日用水量的变化。最重要的变量是建筑年龄、楼层数、每层楼和每人的总面积、