Stochastic Process vs Time Series

Stochastic Process

A stochastic process $\{X_t\}_{t \in T}$ is a collection of random variables indexed by $T$ (the time index). In this course, $T = \{0, 1, 2, \ldots\}$ (discrete time).

Notation

  • $t$ — time index
  • $X_t$ — random variable at time $t$ (the state of the process)

Time Series

A time series is a sequence of observations $x_1, x_2, \ldots, x_n$ — one realization of an underlying stochastic process.

Stochastic ProcessTime Series
NatureTheoretical model (random variables)Observed data (numbers)
Notation$\{X_t\}$ (capital)$\{x_t\}$ (lowercase)
RoleWhat we modelWhat we measure

We only ever observe one realization. To estimate parameters, we need stationarity — treating observations at different times as if from the same distribution.