Description
(a) Build a multiple regression model for the purposes of predicting calories, using all the other variables as the predictors.
(b) Which of the predictors probably does not belong in the model? Explain how you know this. What might be your next step after viewing these results? (Hint: use t-test.)
(c) Suppose we omit cholesterol from the model and rerun the regression. Explain what will happen to the value of R^2.
(d) Which predictor is negatively associated with the response? Explain how you know this. (Hint: Plot and calculate correlation coefficients)
(e) Build a multiple regression model for the purposes of predicting calories, using the forward selection method.
(f) Build a multiple regression model for the purposes of predicting calories, using the backward elimination method.
(g) Build a multiple regression model for the purposes of predicting calories, using the stepwise variable selection procedure.
(h) Apply the best subsets procedure, and compare against the previous methods. (15 points)
(i) Perform all possible regressions. Did the variable selection algorithms find the best regression? (15 points)
Unformatted Attachment Preview
Purchase answer to see full attachment