This paper presents relevant information to understand and apply
linear regression analysis for application on the residential sector with focus on whole-building
energy consumption in single-family homes. The energy signatures and conditional demand analysis
are also discussed for better understanding of the use of practical applications of regression
analysis for residential energy consumption prediction. The literature review of papers dealing
specifically with residential energy consumption using regression analysis supports the feasibility
of this statistical approach for model development. The basics of simple and multiple linear
regression analysis were applied to data from the TxAIRE Research and Demonstration House #1, as a
case study, to illustrate an example of results from regression analysis. As illustrated from the
results of the case study, as the time interval of the observed data increases, the quality of the
models improves. This is explained by the fact that for longer time periods, the discrepancies
among individual effects in shorter time periods are averaged over longer time periods. The
solar radiation as a second predictor variable shows improvement of the coefficient of
determination, but deteriorates the root mean square error, which justifies the importance of using
both parameters to assess the quality of the model based on the developer's criteria. Since HVAC
systems accounts for a large portion of the total energy consumption of buildings, and because the
performance of HVAC systems can be modeled as a second order polynomial, a quadratic regression
model can offer better results for shorter time intervals such as an hour. This is not necessarily
true for longer time periods such as a day because the quadratic trend of the HVAC system
is lost.