For the New York ozone data of Problem 15.2 in Chapter 15, compute stepwise regression of natural log of ozone (i.e., lnOzone) versus the five X variables using the stepwise procedure in R as described in Example 17.2 of Section 17.3. (A) Is the model selected by stepwise regression the same as the reduced model determined in Problem 15.3? (B) Perform the usual checks and diagnostics including VIF tests to verify the absence of multicollinearity, residuals plots to assess compliance with regression assumptions, and the quantitative Shapiro–Wilk, Breusch–Pagan, Durbin– Watson, and Breusch–Godfrey tests for normality, homoscedasticity and independence of the residuals or errors respectively (see Sections 15.4 and 15.5 for descriptions). (C) If the errors are determined to be heteroscedastic and/or autocorrelated, use the HC or HAC robust standard errors as described in Sections 15.5.3 and 15.5.4 to determine whether the coefficients are still significant at the 0.05 significance level. (D) Use the influence.measures and influenceIndexPlot R functions as described in Section 15.9 to check for the presence of influential outliers that might have unduly influenced the regression results. If any influential data points are identified, try removing those observations and recomputing the regression to assess their actual effects on the regression results. (E) Choose any values for the three X variables that are within the range of values in the data sample, and predict the ozone concentration for the specified X values using your regression equation. Is the predicted ozone concentration the mean or median value for the given X values?