14. To dummy-code a categorical variable "University" with the three levels "Western", "Waterloo", "UofT", you decide to use the following coding scheme: X1 X2 Western 0 Waterloo 0 UofT -1 -1 The data are then modelled using the regression model: Y'n = bot bixIn+ b2x2.n You estimate the parameters using a squared-error loss function (OLS regression). Assuming you have equal numbers of samples from each University, what values will the regression coefficients take? O bo: Mean(UofT), b] : Mean(Western)-Mean(UofT), b2 : Mean(Waterloo)- Mean(UofT) O bo: Mean of all data, b, : Mean(Western) - Mean(UofT), b2 : Mean(Waterloo)- Mean(UofT) X bo: Mean of all data, b, : Mean(Western)-Mean(overall) b2 : Mean(Waterloo)- Mean(overall) O bo: Mean of all data, b1 : Mean(Western) b2 : Mean(Waterloo) O bo: Median for UofT, b, : Median(Western)-Median(UofT) b2 : Median(Waterloo)- Median(UofT) O bo: Median of all data, b, :Median(Western)-Median(UofT) b2 : Median(Waterloo)- Median(UofT)