2025 Orange County Water Demand Projection Model
Expectations about
Log Transformed?
Explanatory Variable
Description
Coefficient Estimates
Economic theory suggests positive correlation of income with demand; generally geographical areas with higher median incomes tend to use more water. Positively correlated with demand; generally, residences with more people tend to use larger amounts of water
Positive sign (commonly between 0 and 1)
Median income
Yes
Positive sign (commonly between 0 and 1)
Persons per household
Yes
Reflects the effect of drought restrictions
Water Use Restrictions
No
Negative sign
Most explanatory variables are log-transformed because some variables are orders of magnitude larger than others (for example, income is in the tens of thousands of dollars while monthly precipitation may be smaller than 1 inch). Log transformation compresses the large values and spreads out smaller ones, balancing the data and facilitating the regression’s ability to interpret how each variable affects demand.
3.3 Single-family Regression Development
This section reviews the development of the statistical regression for the single-family residential sector.
3.3.1
Explanatory Variables and Fitted Coefficients
The fit for the final single-family regression is presented in Table 3-3 . While all coefficients are generated with the panel regression approach, Hazen relied on experience and the model fitting process to allows some coefficients to vary by agency (shown as a range of values) and restricted some to be constant across all agencies (and a single value is shown in Table 3-3). Coefficient estimates are within the expected range for all explanatory variables).
3-7
Appendix G - 44
Powered by FlippingBook