Assume we are doing a k-nearest neighbor (kNN) algorithm, and are just using the 2 nearest neighbours (i.e. 2-NN). Each training example has 3 attributes (X1...X3) and the value that is being predicted (Y), as shown in the table below.

X1 X2 X3 Y

10 7 8 15

9 6 5 21

We are given the example (10,5,7) and want to predict the Y value. That is, we have a vector with X1 = 10, X2=5, and X3=7 and want to predict the Y value

Assume that we want to weight the 2NN examples for the prediction of Y using the inverse distance squared as the weighting. What is the predicted value of Y ? Show all workings