Given the following training data for regression (f(x) ) and classification (Class( x) ) tasks. a) Build a decision tree for regression (i.e., a regression tree; ff. slide 32 of the decision tree lecture notes). You may assume that the possible splits are x=1.5,0.5,+0.5,+1.5, +2.5,+3.5,+4.5,+5.5 and the impurity after split (cf. slide 30 of lecture notes) is measured with respect to mean absolute error (MAE) instead of information for classification tasks. In other words, you may simply use MAE to replace information measure and expected MAE (i.e., probability weighted MAE) to replace entropy (i.e., expected information) for our regression task. Show your work for a regression with only 3 levels (i.e., root node level, first intermediate node level and leaf node level). What are the predicted f(x) values for x=0.2 and 3.4 ? b) Following part 1(a), what is the minimum number of levels to obtain a regression tree with zero MAE (i.e., perfect fitting) for the given 9 training data records? Count the leaf node as a level. c) Now, you are asked to build a decision tree for classification (i.e., predicting Class (x); cf. slide 14 of the decision tree lecture notes). You may refer to the possible split values in part 1(a) and the entropy for impurity after split measure. Again, build a decision tree with only 3 levels. What are the predicted Class( x ) values for x=0.2 and 3.4