TD_PACF provides insight as to whether the function being modeled is stationary or not. The partial auto correlations are used to measure the degree of correlation between time series sample points. The algorithm removes the effects of the previous lag. For example, the coefficient for lag 4 focuses on the effect of activity based only at lag 4, with effects of lags 3, 2, and 1 removed. TD_PACF is also used to identify the order of an autoregressive (AR) model, which is a type of time series model that predicts future values of a series based on its past values.
TD_PACF is useful for time series modeling. Some applications of partial autocorrelation include:
ARIMA modeling: Use the TD_PACF plot o determine the order of the autoregressive (AR) term after accounting for the effects of any intervening time steps.
Structural time series modeling: Estimate the relationships between different components of a time series, such as trend, seasonality, and cycle.
Multivariate time series analysis: Analyze the relationships between different time series in a multivariate time series model after accounting for the effects of any intervening time steps.