Question 3: Given the following MLR: > m3.1 vif(m3.1) bathrooms GVIF Df 1 2.796150 1.984903 1 3.009585 1 bedrooms rooms total.living.area 5.235409 1 BasementSize Garage land. acres 2.363039 2 2.457761 1 1.103754 1 main.living.area 3.288041 1 AGE total.full 2.133691 1 2.342271 1 lambda > inverseResponsePlot (m3.1) RSS 1 2.117874 317.8787 2 -1.000000 615.5417 3 0.000000 530.0055 4 1.000000 469.6041 Transforming some of the numerical predictors of MLR m3.1: > summary(powerTransform (cbind(logprice, rooms, land. acers)~1, data=Housing)) bcPower Transformations to Multinormality Est Power Rounded Pwr Wald Lwr Bnd Wald Upr Bnd logprice rooms land. acres 0.4169 0.0478 -0.8070 0.42 0.00 -0.81 0.3965 -0.0176 -0.8414 0.4372 0.1132 -0.7727 A) i) Based on the VIF output, what kind of MLR assumption is violated? What do you suggest to do to fix it? ii) If we consider rooms as a response variable: What would be the value of R of the MLR using the rest of the predictors to predict rooms? "Note": logprice variable is not used. B) What is the suggested for the response variable using Box-Cox transformation (Power Transform) and what is the suggested based on the Inverse Response Plot technique? A based on the inverse response = A based Box - Cox = C) What is the criteria for selecting the best as using: i) Box-Cox Transformation: ii) InverseResponsePlot:_ D) Write down the mathematical formulas of the MLR models as it should be inputted in R- Studio after using the appropriate as for each of transformed predictors and transformed response based on the Box-Cox transformation method.