In a typical scenario, we have an outcome measurement, usually categorical (such as heart attack / no heart attack) or quantitative (such as a temperature measurement), that we wish to predict based on a set of features.

Classification: when we predict categorical (qualitative) outputs

Regression: when we predict quantitative outputs.

Regression

Statistical method that attempts to determine the relationship between the dependent variable $Y$ and a set of independent variables $X$.

• Dependent or response variable: the variable whose values change as a consequence of changes in other values in the system
• Independent or predictors or explanatory variables: regarded as inputs to the system and may take on different values freely.

Widely used for predicting the next value of the dipendent value $Y$ from the values of the independent variables $X$.

Regression VS Forecasting:

• Regression: determine the relationship between two time series
• Forecasting: predict the next value of a time series from the values of one or multiple time series in a dataaset using the learned relationship.

Time Series Forecasting

Univariate vs Multivariate TS

• Time series can be classified as univariate and multivariate.
• It is an important distinction as the selection of the forecasting method depends, among other factors, on this.
• Univariate time series is observation of one variable over time.
• Multivariate time series is a collection of several univariate time series where a series can impact others.