Premium
Suburban Water Demand Modeling Using Stepwise Regression
Author(s) -
Brekke Levi,
Larsen Milton D.,
Ausburn Mary,
Takaichi Lynn
Publication year - 2002
Publication title -
journal ‐ american water works association
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.466
H-Index - 74
eISSN - 1551-8833
pISSN - 0003-150X
DOI - 10.1002/j.1551-8833.2002.tb09558.x
Subject(s) - regression analysis , demand forecasting , regression , stepwise regression , computer science , demand management , linear regression , operations research , econometrics , statistics , engineering , economics , machine learning , mathematics , macroeconomics
Any utility facing rapid growth and expensive capital improvements needs accurate water demand forecasts for planning. For a smaller suburban utility, the planning effort is constrained by limited staff and financial resources. Although smaller utilities may understand the benefits of using multiple regression for water demand forecasting, they may instead decide to use conventional trend analysis or unit water demand analysis. This decision is often based on the belief that multiple regression is difficult to implement, requires staff experts on statistics, and is too time‐consuming. This study was undertaken to eliminate these misperceptions. It shows that: regression‐based water demand models are easy to develop using features available in most spreadsheet and statistical software packages; and, the time to develop a model can be reduced from days to minutes if a systematic model construction procedure such as stepwise regression is used. Although good information is available on using multiple regression analysis to investigate historic water demand, little information is available on the merits of stepwise regression and/or other systematic model construction techniques. This article can help make regression‐based water demand analysis more accessible to water suppliers with limited resources and staff expertise.