AI & ML

Question 1

What is the difference between Supervised and Unsupervised Machine Learning?

Answer 1

Supervised Learning - Utilizes labeled input and output data.

Unsupervised Learning - Discover hidden patterns in data without any human provided labels.

Question 2

What are some of the benefits and drawbacks of Supervised and Unsupervised machine learning when compared to one another?

Answer 2

Supervised Learning Features:
Tend to be more accurate than unsupervised models
Require historical data, or humans to manually label data.

Unsupervised Learning Features:
Do not predict, they simply group data together.

Question 3

Within supervised machine learning models, what are features vs labels?

Answer 3

Supervised Machine learning models “learn” the association between known features and unknown labels.

Each column of data that will help us determine the outcome (win or loss for a tournament game) is called a feature.

The column of data that you are trying to predict is called the label. Machine learning models “learn” the association between features to predict the outcome of a label.

Question 4

If we were training a classification supervised machine learning model (e.g. a logistic regression) on historic team sports results to predict the outcome of future games, why should we NOT use the points scored (win_pts or lose_pts) as a feature in our training dataset, even though we have the data available?

Answer 4

This feature is only available at the END of the game and for future games we are making predictions before a game begins.