DEVELOPMENT OF TIME SERIES MODELS TO FORCAST THE RETAIL PRICES OF FOOD ITEMS IN GOMBE STATE,NIGERIA

Abstract

Models for both linear and quadratic trends for food items in Gombe state have been
obtained. Correlogram and Partial autocorrelation of the time series data are plotted in
which the correlogram decays rapidly to zero, indicating that the series is stationary. Also,
the partial autocorrelation cuts off after lag 1 suggesting an autoregressive (AR) model of
order 2. The models obtained were used in making the forecast. It has been observed that
the quadratic trend is a much superior than the linear one. Recommendations are made
which include; the Government should ensure the accurate generation of data, storage of
same, retrieval and use for proper planning.