The file P03_65.xlsx contains basketball data on all NBA teams for five seasons. The SRS (simple rating system) variable is a measure of how good a team is in any given year.
a. Given the explanation of SRS, it makes sense to use multiple regression, with PTS and O_PTS as the explanatory variables, to predict SRS. Do you get a good fit?
b. Suppose instead that the goal is to predict Wins. Try multiple regression, using the variables in columns G–AH or variables calculated from them. For example, instead of FG and FGA, you could try FG/FGA, the fraction of attempted field goals made. You will have to guard against exact multicollinearity.
For example, PTS can be calculated exactly from FG, 3P, and FT. This is a good time to use some form of stepwise regression. How well is your best equation able to predict Wins?