Regression (Data Science)
- •
def
- •
predict continuous values
- •
- •
e.g.
- •
price prediction
- •
- •
Purpose
- •
predict a numerical response and study how it varies with supplid features
- •
- •
Methods
- •
Linear Regression is a basic model; Unavailable reference is a common fitting criterion
- •
- •
Evaluation
- •
evaluate predictions on unseen observations using an appropriate measure such as Root Mean Square Error
- •
- •
Caution
- •
predictive relationships do not establish causation withotu an appropriate design and asusmptions
- •
- •
Distinguish from
- •
Unavailable reference usually serves classification despite its name
- •