Bias/Variance/Over-fitting

Question 1

Bias

Answer 1

how far a model’s predictions are from the target

Question 2

Variance

Answer 2

how your model reacts to changes in the training data

Question 3

As the model becomes more complex

Answer 3

it picks up patterns in the training data making it less generalizable

Question 4

A model with many predictive attributes will exhibit

Answer 4

low bias, high variance

Question 5

A model with too few predictive attributes will exhibit

Answer 5

low variance, but may be quite biased

Question 6

bias informally

Answer 6

how far a model’s predictions are from target (underfitting)

Question 7

variance informally

Answer 7

the degree to which these predictions vary between model iterations (overfitting)

Question 8

complex models have

Answer 8

higher variance

Question 9

Dimensionality reduction and feature selection can

Answer 9

reduce variance by simplifying models

Question 10

Regularization can help

Answer 10

reduce variance

Question 11

If you cannot reduce dimensions or engage in feature selection, what can help decrease variance

Answer 11

A larger training set