Problem 4: k-Nearest Neighbors and Decision Trees [6 points] Consider this training dataset, where each example has two features, X1 and X2, and a label, Y: Training Set 12 Y=0 Y=1 10 6 4 N 0 -2 -500 0 500 1000 1500 2000 2500 X1 Will K-NN classification work well if we apply it directly (without any data pre-processing) to this training dataset? Argue why or why not. If not, how would you pre-process the data to address the issue? Will decision-tree classification work well if we apply it directly? Draw a decision tree with two decision nodes (specifying the feature variables and their thresholds), which fits the training data best, and estimate its accuracy when predicting the training examples