What is autocorrelation function (ACF) and how is it used in time series modeling?

ACF measures correlation between a time series and its lagged versions to identify temporal dependencies.

The Autocorrelation Function (ACF) quantifies the correlation between a time series and its lagged values across different lag distances. It helps identify repeating patterns and is crucial for selecting MA(q) order in ARIMA models. ACF plots show how correlation decays over time; slow decay indicates non-stationarity, while sharp cutoff suggests MA behavior.