MODEL COMPARISON OF VECTOR AUTOREGRESSIVE RESHAPED AND SARIMA IN SEASONAL DATA (A CASE STUDY OF TEA PRODUCTION IN PT PERKEBUNAN NUSANTARA VIII INDONESIA)

Abstract

PT Perkebunan Nusantara VIII (PTPN VIII) is a State-Owned Enterprise (BUMN). It operates in the plantation sector.  The leading commodity is tea.  The demand for tea produced by PTPN VIII is increasing. Thus, planning tea production is necessary. One of the production planning efforts is through forecasting based on previous data.  Tea  production data is time-series data.  It contains seasonal elements and is dependent on other locations. We will analyze data with these criteria  using space-time models, one of which is vector autoregressive (VAR). VAR models the relationship  between observations on certain variables at one time. It also models the observation of the variable itself at previous times. Additionally, VAR models  the relationship  between observations and other variables at previous times. This paper explains how to forecast tea  production. It uses the reconstituted VAR and Seasonal Autoregressive Moving Average (SARIMA) models. The results showed that the reconstituted VAR model was better than the SARIMA model in predicting tea production. The tea production prediction was at the Sedep and Santosa plantations in Bandung Regency.