This problem requires statistical software. The complete data set on energy consumption given for Exercise 7 in Chapter 7 contains information on other variables that may affect power consumption. Table 8.23 includes:
TMAX: maximum daily temperature
TMIN: minimum daily temperature
WNDSPD: coded “0” if less than 6 knots and “1” if 6 or more knots
CLDCOVER: on an ordinal scale from 0 to 3
KWH: electricity consumption
(a) Build a regression model that uses all these independent variables to predict KWH.
(b) Examine the residuals graphically for evidence of curvilinearity or heteroscedasticity.
(c) Examine the outlier diagnostics for possible outliers.
(d) Examine the data for multicollinearity.
(e) Is there a simpler model (using fewer independent variables) that will work nearly as well as the full model?
(f) Using your simplest adequate model, interpret the contribution of the independent variables to energy consumption.
Exercise 7
In an effort to determine the cost of air conditioning, a resident in College Station, TX, recorded daily values of the variables
Tavg = mean temperature
Kwh = electricity consumption
for the period from September 19 through November 4 (Table 7.19).
(a) Make a scatterplot to show the relationship of power consumption and temperature.
(b) Using the model
Kwh = β0
+ β1
(Tavg) + ε;
estimate the parameters, test appropriate hypotheses, and write a short paragraph stating your findings.
(c) Make a residual plot to see whether the model appears to be appropriately specified.