• 1) explain the differences between knn classification and knn regression. <- want a better answer. i know difference between classification/regression, but don't know how to specifically tie it into knn 2
• ) the loess model and knn are both local methods that require neighborhood definitions. comment on the bias-variance trade-off when defining a neighborhood for knn regression. <- just said they're more flexible and therefore more variance etc 
• 3) if the bayes decision boundary is linear, do we expect linear discriminant analysis (lda) or quadratic discriminant analysis (qda) to perform better on the training data set? on the test data set? <- i saw an answer for this online, but i decided to use my own and try to put it into my own words rather than plagarize, but going through iut more thoroughly would be good consider a two-class classification problem y ∈ {0, 1} with one predictor x, where the pr(y = 1) = 0.7. the distribution of x for an observation that comes from class 0 is a standard normal, given by the conditional density f(x|y = 0) = f0(x) = got the equation down already  . the distribution of x in class 1 is also normal, but has mean 1.5 and variance 0.5. .
• (a) find the bayes decision boundary. - got it. it was quadratic, approximately 0.375 and 0.544 
• (b) if we had a representative sample from this population, would we expect the lda or qda decision boundary to be closer to Bayes decision boundary? why?