so i could not understand if the last significant pacf value ( above the threshold is lets say 4(months)) we will discard all the values after 4 month from the pacf equation . But why don't we discard the value below 4 which are not statistically significant from the Pacf plot
Hi Jagteshwar,
So, the last significant PACF value indicates the point beyond which the lagged correlations are statistically insignificant. Values beyond this lag are generally excluded because they contribute minimal or no additional explanatory power to the model. Including them could overfit the data or introduce unnecessary complexity.
But we keep all the lags before it, even if some seem insignificant, to maintain the model’s structure. Skipping those earlier lags can mess with how the relationships between lags are accounted for and might bias the results. Plus, those insignificant lags often play a role in explaining the data, even if it's subtle, so it's better to include them for a more robust model.
Hope this helps!