Multivariate Time Series Model for Forecasting Urad Price in Different Zones of India

摘要

A natural extension of univariate autoregressive model to dynamic multivariate time series is called Vector autoregressive (VAR). Vector autoregressive is one of the most successful, flexible models in multivariate time series. This study is an attempt to review the theory and applications of VAR model based on historic data. VAR model is applied to the monthly retail price data of uradin five major regions namely north zone (NZ), west zone (WZ), east zone (EZ), north east zone (NEZ) and south zone (SZ)of India for forecasting. The best model is selected using Akaike information criterion and Schwarz information criteria. Forecast evaluation is carried out in terms of Relative mean absolute percentage error (RMAPE)and Root mean square error (RMSE).