Machine Learning

Question 1

Machine learning

Answer 1

Algorithms that can learn from observational data and make predictions from it

Question 2

Unsupervised learning

Answer 2

An algorithm makes sense of a data set without prior learning experience or answers to learn from

Question 3

Latent variable

Answer 3

A previously unknown part of the data, which unsupervised learning can do

Question 4

Supervised learning

Answer 4

An algorithm learns from a data set plus the correct “answers”

Question 5

Training/testing sets

Answer 5

A model is trained using a training set of data, then the model is tested on a similar but disjoint set of data to test its accuracy.

Question 6

What are practical considerations for training/testing sets?

Answer 6

1. Both sets must be large to have many outliers and variations.
2. Both sets must be randomly chosen from the source data pool.

Question 7

Why is train/test useful?

Answer 7

It can guard against overfitting.

Question 8

K-fold cross variation

Answer 8

1. Split data randomly into K segments.
2. Take one segment as the test set.
3. Train on the other sets and compare with the test set.
4. Average the resulting r-squared values.

Question 9

K-means clustering

Answer 9

1. Randomly pick K centroids.
2. Assign each data point to the closest centroid.
3. Recompute the centroids based on the average position of each centroid’s data points.
4. Iterate until points stop moving.

Question 10

What is a large caveat with K-means clustering?

Answer 10

The algorithm does not assign names or titles to clusters.

Question 11

Entropy (data science)

Answer 11

Disorder of data

Zero if all data points are the same.