PACF (Partial Autocorrelation Function)

The Partial Autocorrelation Function (PACF) measures the direct relationship between a data point and its past values, removing the influence of intermediate points. In simple terms, it tells us how strongly a specific lag (like 2 days ago) predicts the current value, considering all the shorter lags in between. This helps identify the actual number of past points influencing the present, which is useful for building accurate time series models, such as AR models. PACF helps distinguish between genuine long-term relationships and those caused by shorter-term correlations.