Multiple Regression II

Question 1

Interaction

Answer 1

-Effects are dependent on each other
-Response to a treatment depends on sex
-Different from additive effect because the magnitude of the effect of the variable on the outcome variable depends on another variable

Question 2

Bias-variance tradeoff

Answer 2

Model can’t have both low bias and low variance, so there is a “tradeoff” between bias and variance

Question 3

Underfitting

Answer 3

-High bias and low variance
-Not learning enough from the data and can’t capture the underlying trend of the data
-Not complex enough

Question 4

Overfitting

Answer 4

-High variance and low bias
-Too tailored for your data set
-Too complex

Question 5

To address under/overfitting…

Answer 5

Score the model given the model’s complexity using cross validating and information criteria

Question 6

Information criteria

Answer 6

-Estimates out of sample deviance
-Lower value means less prediction error and lower out of sample deviance