Question

8. Decision Tree Learning Output Alt Yes Yes No Yes Fri No No No $ No Bar No No Yes No No No NO Yes Yes Yes Yes No Yes No $$$ No No Yes Yes Yes HunPatPrice Rain ResType Yes Some $$$ French Yes Full No Thai No Some No Burger Yes Full No No Thai Full Yes French Yes Some Yes Yes Italian No None Yes No Burger Yes Some Yes Yes Thai Full No Burger Yes Full No Yes Italian No None No Thai Yes Full $ Burger Est 0 - 10 30 - 60 0 - 10 10-30 > 60 0 - 10 0 - 10 0 - 10 > 60 10 – 30 0 - 10 | 30 - 60 No No No Z3333 Yes No Yes No Yes Yes No Yes Yes Yes No $$$ No No No Yes No Yes No Given the 12 training examples in the table, above, which feature would be tested first in a decision tree trained to maximize information gain at each test?

Answer 1

will give consistent prediction confidence since the basic logic behind it is to explore the neighborhood and assumes that the point would behave the same way and to determine we put an odd number as the k value. Its advantages is its easy and simple to machine learning model to implement and few hyperparameter to tune. Some disadvantages can be to get the k value correct, it takes a lot of time to compute if the dataset is huge and there should be proper scaling among features so that it works properly. The two hyperparameters that has to be taken care is the k value and the distance measurement -like euclidean, hamming, manhattan or minkowski.

b) When we know a particular column can effect the prediction column hugely then we can use linear regression or choosing from the given models we can use

But all in all, the data-set has to be study and feature selection is done, data is cleaned many steps are preformed such as data preproccesing and then different models are tried and then we see which preforms better and then that is selected.