Aeroplan, the loyalty program of Air Canada, has been in operation since 1984. Travellers on Air Canada flights receive Aeroplan miles, which can be redeemed for free travel, vacations, or merchandise. Aeroplan was originally owned by Air Canada and was spun off as a separate company in 2002. Its initial public offering in 2005 created the first publicly traded loyalty company worldwide. In 2011 it changed its name to AIMIA, derived from the word "aim", i.e. targeting the right customers. Today, AIMIA manages loyalty programs and provides loyalty analytics for many companies in a broad range of industries. AIMIA donates Aeroplan miles to international organization, such as Engineers Without Borders and War Child Canada. Some of the miles donated come from AIMIA itself, matching donations by Aeroplan members. To encourage members to make donations, loyalty programs track certain information about those people who do make donations. If this information is related to the number of miles donated, then the loyalty program could target advertising to members who are expected to donate a large number of miles. In the case of frequent flyer miles, the following information would probably be available: D = Number of miles donated during the past five years I = Average household income of the member's postal code N = Number of loyalty miles earned during the past five years S = Percentage of flights taken during the past five years on which a special meal was ordered (limited to those flights on which special meals were available) Suppose the file

https://docs.google.com/spreadsheets/d/14rXHSB00wV7vdlUkNc96AC8KFM3nMlPXWvWZ-A15Bac/edit?usp=sharing contains data from a random sample of frequent flyers who donated loyalty miles.

a) Using Minitab, provide a regression analysis report (Regression Equation, Coefficients, Model Summary, Analysis of Variance, Unusual Observations). What is the regression equation in which D is estimated as a function of I, N, and S? b) Using Minitab, using either Best Subsets or Stepwise Regression approach (backward elimination method), develop the best multiple regression model. Use a 5% significance level. Explain your results.