Lecture 6 - Machine Learning

Question 1

What is the training process in machine learning?

Answer 1

The training process is an iterative optimization problem. This leaves us with the “best” model we can get.

Question 2

How do we measure a classification models performance?

Answer 2

Accuracy
Precision
Recall
F-1 Score

Question 3

How do we measure a linear regression models performance?

Answer 3

Mean Squared Error (MSE)

R2 score

Question 4

What is accuracy?

Answer 4

Correct / total

Question 5

What are some weaknesses of accuracy?

Answer 5

Does not discriminate between type I and type II errors, does not handle class imbalance

Question 6

What is precision?

Answer 6

True positives / Truepositives + Falsepositives

It is the fraction of relevant instances among the retrieved instances.

Question 7

What is recall?

Answer 7

True positiveness / truepositive + falsenegative

Recall is the fraction of relevant instances that have been retrieved over the total amount of relevant instances.

Question 8

What is F-1 Score?

Answer 8

2 (( Precision * recall) / (precision + Recall)

Pros: It’s a single number that captures a lot of detail Cons: It’s a single number that captures a lot of detail, so some is lost or “glossed over”

Question 9

What is a Type I error?

Answer 9

A false positive

Question 10

What is a Type II error?

Answer 10

A false negative

Question 11

What is overfitting?

Answer 11

Abstractly, overfitting is when you come close to “memorizing” your data set. It means you perform very well on the training data set, but your model doesn’t generalize well to unseen data