In: Statistics and Probability
Solve it by R
Use the ‘cement’ dataset in ‘MASS’ package to answer the question.
(1) Conduct the multiple linear regression, regress y value on x1, x2, x3 and x4 (without intercept). Report the estimated coefficients. Which predictor variables have strong linear relationship with response variable y at significance level 0.05?
(2) What is the adjusted R square of your regression? What is the interquartile range (IQR) of the residuals from your regression?
(3) Conduct a best subset regression (with intercept) with the function ‘regsubsets’ and find the best model with only two independent variables.