This question in R Programming

Here we want to create a 5-NN classifier of the iris data. Following the discussion in class, for each instance in

our data set we will find the 5 closest instances (other than the instance itself). It may help to know about the

order function in R: If x is vector of numbers, order(x) gives the order of elements. So if index=order(x) then

index[1] is the the index of the smallest element, index[2] is the index of the 2nd smallest, etc. Said another

way x[index[1]] is the smallest value of x, x[index[2]] is the 2nd smallest, etc. Use this information to create a

5-NN classifier for the iris data. That is, your classifier should loop through each row of the distance matrix

and tally the class labels of the 5 closest instances (other than the instance itself). You would classify the

instance according to the most prevalent class within the 5 neighbors, breaking ties arbitrarily.

(a) Compute a vector, classhat, which is the result of your classifier applied to each instance.

(b) Compute the error rate of your classifier.

(c) Explain why you do or dont believe your error rate is an accurate estimate of the generalization error

rate.