# Intro

* Predictive or Supervised: learn a mappping from inputs x to outputs u, given a labeled set of input-output paris (the training set).
	- The training input x_i is called features, attributes, covariates.
	- If y_i assumes a value from a finite set, it's called categorical or nominal, and the problem is classification or pattern recognition. If y_i us real-valued scalar, it is regression.

* Descriptive or unsupervised learning: find patterns in the data (knowledge discovery). c