(3.a) For KNN with K neighboring observations for the purpose of regression, i.e., for the purpose of predicting the random variable y using another X, what does the estimate provided by KNN approach to when K is very large? (3.b) For using a function f of a random variable X to predict another Y, what is the relationship between the minimizer of the integrated squared loss and the conditional expectation of Y given X? Do you see a connection between your answers to (3.a) and (3.b)? (3.a) For KNN with K neighboring observations for the purpose of regression, i.e., for the purpose of predicting the random variable y using another X, what does the estimate provided by KNN approach to when K is very large? (3.b) For using a function f of a random variable X to predict another Y, what is the relationship between the minimizer of the integrated squared loss and the conditional expectation of Y given X? Do you see a connection between your answers to (3.a) and (3.b)