An Autoregressive Moving Average model \( \text{ARMA}(p, q) \) combines autoregressive (AR) and moving average (MA) terms. In its most common form, this model is described as:

\( Y_t = c + \phi_1 Y_{t-1} + \cdots + \phi_p Y_{t-p} + \epsilon_t + \theta_1 \epsilon_{t-1} + \cdots + \theta_q \epsilon_{t-q}, \)

where \( \{\epsilon_t\} \) is a white noise process with zero mean and constant variance.

For the model to be valid, the series must not have a unit root; in other words, it must be wide-sense stationary.

These models are highly useful for modeling and forecasting time series that exhibit linear dependencies, and they are widely applied in fields such as economics, engineering, and natural sciences.