Hybrid singular spectrum analysis-ARIMA modelling for direct and indirect forecasting of Farmer's Term of Trade in East Java

Agricultural sector has a significant contribution to the economy of East Java. The important role of agricultural sector in East Java needs good planning so that development process can be done as expected goals. Farmer's Terms of Trade (FTT) is one of the indicators that can be used to measure the success of development in the agricultural sector. Therefore, appropriate FTT modelling will be very useful to determine policies in the agricultural sector. SSA is a time series data analysis that aims to decompose the original time series into the sum of its component, i.e. trend, oscillatory, and noise. In previous studies, SSA could provide better forecasting results compared with other methods when used in data with the complex structure such as data with more than one seasonal component. FTT is a time series data composed of several data series with diverse characteristics and more than one type of seasonal component period. It makes the data structure of FTT become complex. In this research, hybrid SSA-ARIMA method is used in FTT forecasting directly and indirectly. The result shows that direct forecasting of FTT using SSA-ARIMA method gives the better result than ARIMA. While on indirect FTT forecasting, ARIMA yields better result than SSA-ARIMA. It shows that SSA-ARIMA provides the better results than ARIMA if applied in complex data. In general, in some FTT forecasting methods and approaches, direct FTT forecasting using the SSA-ARIMA method provides the best forecasting results.