Understand the fundamentals of decision tree. For classification problems, given a simple

dataset, know $(1)$ how to calculate entropy, $(2)$ how to compare different splits using

information gain, $(3)$ how to construct a decision tree using entropy$-$based information gain.

Example:

Question $2\mathrm{.}1$: Given the dataset above, draw a decision tree using entropy$-$based information

gain such that it can classify the Target for all data samples without any errors.

Question $2\mathrm{.}2$: What prediction will your decision tree make for the following query?

Head Shape $=$ square, Body Shape $=$ rectangular, Body Color $=$ black Understand the concept of bootstrapping$/$bagging$,$ and how ensemble learning work. Given a

simple dataset, know how to construct a random forest model.

Practice question: consider the task of predicting whether a post$-$operative patient should be sent

to an intensive care unit $($ICU$)$ or to a general ward for recovery. The dataset below contains the

details of six patients in a study. CORE describes the core temperature of the patient $($which can

be low or high$)$ and STABLE describes whether the patient's current temperature is stable $($true

or false$).$ The target feature in this domain, DECISION, records the decision of whether the

patient is sent to the icu or to a general ward $($gen$)$ for recovery. The goal is to build a random

forest model for this post$-$operative patient routing task.

Question $3\mathrm{.}1$: the three tables below list three bootstrap samples that have been generated from

the above dataset. Using these bootstrap samples create the decision trees that will be in the

random forest model $($use entropy$-$based information gain as the feature selection criterion$).$

Bootstrap Sample A:

Bootstrap Sample B: