Lesson 2

Question 1

What is the exhaustive search?

Decision trees

Answer 1

Trying every possible way to split the data set and constructing every possible tree

For every feature the tree grows exponentially

Question 2

What does a high impurity mean?

Answer 2

This means that a node has a more equal distribution

Question 3

How can you calculate the depth of a decision tree?

Answer 3

It is the maximum amount of questions you can ask to get an answer

Question 4

What are 7 advantages of Decision trees

Answer 4

• easy to understand
• can be visualised
• handles both numerical and categorical data
• requires little data preparation
• captures non-linear relationships
• feature importance
• fast to train and predict

Question 5

What are 4 drawbacks of decision trees

Answer 5

• Prone to overfitting
• Sensitive to minor details
• Unstable
• Biased toward features with more levels

Question 6

What is ensemble learning?

Answer 6

When ‘weak learner’ models are combined either sequential or parallel.

“The wisdom of the crowd”

Question 7

What is Bagging?

Decision trees

Answer 7

Using different training sets to reduce variance

Parallel ensemble learning technique

Question 8

What is Random Forest?

Answer 8

Using different training sets and different features to reduce variance