TY - JOUR
T1 - A self-validation method for regression models of unitary HVAC equipment based on manufacturers' data
AU - Yang, Huojun
AU - Li, Haorong
AU - Yuill, David P.
PY - 2013/2/1
Y1 - 2013/2/1
N2 - Regression models of various types are often used to predict the performance of unitary HVAC systems. Models must be validated before they can be used with any confidence. This article describes a self-validation method for regression models based on the widely-available manufacturer's performance data. The self-validation method also uses the manufacturer's data, thus avoiding the problem of scarce or expensive laboratory test validation data. It is derived based on the Fundamental Theorem of Calculus. In this method, the discrete manufacturer's data are studied within the context of the continuous domain, so that the partial derivatives of the regression models derived from the manufacturer's data are approximated for each independent variable. The derivative of a regression model is said to be effectively validated when all of its partial derivatives are valid, and then the quality of the regression model can be further evaluated for its intended use. The self-validation method was implemented to validate models of unitary HVAC equipment that were developed with Generic Rating-Data-Based modeling, and it showed that self-validation using the manufacturer's data is consistent with laboratory data based validation.
AB - Regression models of various types are often used to predict the performance of unitary HVAC systems. Models must be validated before they can be used with any confidence. This article describes a self-validation method for regression models based on the widely-available manufacturer's performance data. The self-validation method also uses the manufacturer's data, thus avoiding the problem of scarce or expensive laboratory test validation data. It is derived based on the Fundamental Theorem of Calculus. In this method, the discrete manufacturer's data are studied within the context of the continuous domain, so that the partial derivatives of the regression models derived from the manufacturer's data are approximated for each independent variable. The derivative of a regression model is said to be effectively validated when all of its partial derivatives are valid, and then the quality of the regression model can be further evaluated for its intended use. The self-validation method was implemented to validate models of unitary HVAC equipment that were developed with Generic Rating-Data-Based modeling, and it showed that self-validation using the manufacturer's data is consistent with laboratory data based validation.
UR - http://www.scopus.com/inward/record.url?scp=84874480133&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=84874480133&partnerID=8YFLogxK
U2 - 10.1080/10789669.2012.754307
DO - 10.1080/10789669.2012.754307
M3 - Article
AN - SCOPUS:84874480133
VL - 19
SP - 175
EP - 192
JO - Science and Technology for the Built Environment
JF - Science and Technology for the Built Environment
SN - 2374-4731
IS - 2
ER -