VECTOR_REGRESSION

Vector Autoregression (VAR) is a multivariate time series forecasting model that predicts multiple interdependent variables simultaneously. Unlike traditional linear regression, VAR treats all variables as endogenous (dependent on each other) and models their lagged values to understand how they influence each other over time.

Advantages of VAR

✔ Handles multiple time series together (e.g., demand & supply, ad spend & sales)

✔ Captures dynamic relationships between variables

✔ No need for external predictors—uses lagged values of all variables

✔ Can be extended to structural VAR (SVAR) for deeper economic analysis