PSTAT 131 TEST PAPER ANSWERS AND QUESTIONS
SET A+
Bayes Classifier... - ✔✔produces lowest test MSE
✔✔Expected test MSE - ✔✔Bias^2+var+irreducible error
✔✔overfitting occurs when - ✔✔flexible method learns too hard
✔✔KNN - ✔✔supervised. As k increases, bias increased and variance decreases
✔✔Curse of Dimensionality - ✔✔failing of KNN when p very large
✔✔In linear regression the training SSR... - ✔✔can be zero
✔✔Least Squares estimates have.... - ✔✔zero bias
✔✔logit(p)= - ✔✔log(p/(p-1))
✔✔A logistic regression is a method where - ✔✔logit of y is a linear function of x
✔✔MSE is not suitable for evaluating.... - ✔✔logistic regression model
✔✔which is more flexible: QDA or LDA? - ✔✔QDA
✔✔QDA assumes - ✔✔different covariances
✔✔LDA assumes - ✔✔same covariances
✔✔QDA and LDA both assume - ✔✔different mean, normal distribution
✔✔ROC curve - ✔✔identify appropriate tradeoff between TPR and FPR, perfect
classifier contains AUC of 1
SET A+
Bayes Classifier... - ✔✔produces lowest test MSE
✔✔Expected test MSE - ✔✔Bias^2+var+irreducible error
✔✔overfitting occurs when - ✔✔flexible method learns too hard
✔✔KNN - ✔✔supervised. As k increases, bias increased and variance decreases
✔✔Curse of Dimensionality - ✔✔failing of KNN when p very large
✔✔In linear regression the training SSR... - ✔✔can be zero
✔✔Least Squares estimates have.... - ✔✔zero bias
✔✔logit(p)= - ✔✔log(p/(p-1))
✔✔A logistic regression is a method where - ✔✔logit of y is a linear function of x
✔✔MSE is not suitable for evaluating.... - ✔✔logistic regression model
✔✔which is more flexible: QDA or LDA? - ✔✔QDA
✔✔QDA assumes - ✔✔different covariances
✔✔LDA assumes - ✔✔same covariances
✔✔QDA and LDA both assume - ✔✔different mean, normal distribution
✔✔ROC curve - ✔✔identify appropriate tradeoff between TPR and FPR, perfect
classifier contains AUC of 1
