Hands-On Machine Learning on Google Cloud Platform

Introducing time series

A time series constitutes a sequence of observations on a phenomenon y carried out in consecutive instants or time intervals that are usually, even if not necessarily, evenly spaced or of the same length. The trend of commodity prices, stock market indices, the BTP/BUND spread, and the unemployment rate are just a few examples of times series.

Contrary to what happens in classical statistics, where it is assumed that an independent observations come from a single random variable, in a time series, it is assumed that there are n observations coming from as many dependent random variables. The inference of the time series is thus configured as a procedure that attempts to bring the time series back to its generating ...