In a supervised learning problem, there is an explicit, pre-defined target variable which has known labels. If the outcome is continuous, the problem is solved via a regression context. If the outcome has discrete categories, a classification approach is taken.

More formally, Supervised Learning refers to the task of learning from ‘Training Data’. In this process, the algorithms learn a mapping from input data to output, by looking at several pairs of input – output (\(\mathbf{x}\), \(y\)) data.

Examples:

*Regression Analysis*

– Predict the selling price of a house

– Learn from the housing data from the past to make prediction/estimates about the current selling price

*Classification Analysis*

– Predict is the stock market with go UP or Down tomorrow

– Learn from past stock market data to map if the present day data warrants an increase/decrease tomorrow.