Keywords: multiple regression, variance modelling.
Sanford Weisberg writes
When gasoline is pumped into a tank, hydrocarbon vapors are forced out of a tank and into the atmosphere. To reduce this significant source of air pollution, devices are installed to capture the vapor. In testing these vapor recovery systems, the amount that escapes cannot be measured, but a "sniffer" can determine if some vapor is escaping. Also, the amount that is recovered can be measured. To estimate the efficiency of the system, some method of estimating the total amount given off must be used. To this end, a laboratory experiment was conducted in which the amount of vapor given off was measured under carefully controlled conditions. Four variables are relevant for modeling. In an experiment, these conditions were varied and the quantity of emitted hydrocarbons was measured in grams.
|TankTemp||-||initial tank temperature (°F)|
|GasTemp||-||temperature of the dispensed gasoline (°F)|
|TankPres||-||initial vapor pressure in the tank (psi)|
|GasPres||-||vapor pressure of the dispensed gasoline (psi)|
|HC||-||emitted hydrocarbons (g)|
Data File (tab-delimited text)
|Weisberg, S. (1985). Applied Linear Regression. Wiley, New York. Example 6.2. John Rice is cited as the original source.|
|Smyth, G. K. (1989). Generalized linear models with varying dispersion. J. Roy. Statist. Soc. B, 51, 47-60.|