Search for question

trips per household as a function of the remaining variables (to the extent possible). For final model

specifications, interpret the significant variables as appropriate. Feel free to be creative with the

variables you create and whether you choose to log-transform specific variables, as variable creation

will be part of the grading consideration.

While fitting your linear regression model, you should complete the following:

a) Estimate a linear regression using a forward step-wise procedure until you reach your final

model specifications.

• Consider if heteroskedasciticy is present and how you should address this.

• Is autocorrelation a concern?

• Remember to present the descriptive statistics for your significant variables (see Canvas

document).

b) Check for correlation among independent variables to ensure there are no collinearity issues.

Correlation matrix (as you are fitting your model).

Variance inflation factors (post-estimation).

e) Check that all significant variables in your final model specification make sense (i.e., does

having a positive/negative effect on Y make sense).

d) Report the number of observations, R², and Adjusted R² in final model specifications.

e) Check if the mean of the residuals is close to zero. Comment on the mean.

f) Plot the histogram of the standardized residuals. Comment on whether the standardized

residuals look normal.

g) Plot the residuals vs. fitted and fitted values vs. actual values. Comment.

h) Interpret all of the variables in your final model specifications and provide plausible explana-

tions for the effects you observe. Discuss the logical process that led you to your final model

specifications.