(A) Explain hold-out and 10-fold CV and their strengths and possible weaknesses. (B) Explain the overfitting problem. (C) Explain random forest method and how it can improve upon the standard decision tree for regression problems. (D) State one obvious reason why linear regression is not appropriate for binary/categorical response variable. (E) Below is the optimisation problem for soft margin classifier (i) (ii) maximize M Bo Bp.1..M subject to E B =1 4- j=1 Yi (30 + 31x1 + + BpXip) M (1-;) n 20.6 C i=1 where C is a non-negative tuning parameter. Explain the role of variable M. Explain the relationships between C and the variables ,i = 1, ..., n, where n is the number of observations in the training dataset.