Week 2

Question 1

What are these: Classification, regression

Answer 1

Supervised learning

Question 2

What are these: Clustering, outlier detection

Answer 2

Unsupervised learning

Question 3

Classification predicts ________ values whereas regression predicts _________ values

Answer 3

discrete values whereas regression predicts continuous values (REGRESSION CAN ALSO TAKE CATEGORICAL VALUES)

Question 4

Name this: The data has no class labels

Answer 4

Unsupervised learning

Question 5

Name this: Data has class labels

Answer 5

Supervised learning

Question 6

T/F - You cannot use the same data to test your algorithm, you need an independent test set

Answer 6

True. (Training set and test set)

Question 7

What is model overfitting?

Answer 7

When you try to get 100% accuracy, this generally means completely tailoring your model to the training data, it will not longer work well on the evaluation data

Question 8

What is model underfitting?

Answer 8

When the model performs poorly on the training data

Question 9

Name 2 types of error estimation

Answer 9

1. Random sampling with repeated holdout
2. K-fold cross-validation

Question 10

T/F - In K-fold cross validation we split the data into (usually) 10 equal groups, train on the (usually) first 9 groups and test on the last group.

Answer 10

True

Question 11

How is the overall accuracy found in K-fold cross validation?

Answer 11

Average accuracy rate = overall accuracy

Question 12

Place these in the correct boxes
TP FN FP TN

 P+ P- A+ A-

Answer 12

TP FN
FP TN