I. 'I'rue/False/Unoertain - Briey explain. No credit without an explanation (8 marks each). 1. With 5-fold cross validation. 20% of the sample data are never used to estimate the model, just to estimate the TMSE. 2. TMSE with cross-validation is not always decreasing when adding more Xe, so it is a guru] estinmtor or GMSE. 3. Using the bootstrap procedure for model averaging, not every observation of the original sample will be selected at least once. 4. One can use the LRT to test the hypothesis Ho : 65 = O in a legit model. 5. The slope of the probit model is maximized when d5 = 0.5