Development of Regression Models considering Time-Lag and Aerosols for Predicting Heating Loads in Buildings

Table 8

Residual statistics for a DOE small office building depending on different models.

Residual statistics^a
		Minimum	Maximum	Mean	Std. deviation	N
Modified model^b	Predicted value	−16410367.00	28590780.00	3786390.10	5224087.118	20,350
	Residual	−20649606.000	64699612.000	0.000	7114276.785	20,350
	Std. predicted value	−3.866	4.748	0.000	1.000	20,350
	Std. residual	−2.902	9.092	0.000	1.000	20,350
New model^c	Predicted value	−2313629.00	13152829.00	1123572.72	1738678.448	4854
	Residual	−10689199.000	46347212.000	0.000	3739487.258	4854
	Std. predicted value	−1.977	6.919	0.000	1.000	4854
	Std. residual	−2.856	12.384	0.000	0.999	4854
Limited model^d	Predicted value	−8359116.00	17571524.00	3786390.10	4531787.978	20,350
	Residual	−15785886.000	70738864.000	0.000	7574095.201	20,350
	Std. predicted value	−2.680	3.042	0.000	1.000	20,350
	Std. residual	−2.084	9.338	0.000	1.000	20,350

^aDependent variable: heating load of a DOE small office building. ^bPredictors: (constant), visibility, diffuse radiation, atmospheric station pressure, wind speed, total sky cover, dry bulb temperatures, direct normal radiation, relative humidity, global horizontal radiation, horizontal infrared radiation, and dew point temperatures; ^cpredictors: (constant), precipitable water, direct normal radiation, total sky cover, AOD, relative humidity, atmospheric station pressure, diffuse radiation, dew point temperatures, global horizontal radiation, and horizontal infrared radiation; ^dpredictors: (constant), wind speed, dry bulb temperatures, relative humidity, global horizontal radiation, atmospheric station pressure, and dew point temperatures.