<![CDATA[Forecasting the number of airplane passengers requires a model related to time and location because the data on the number of airplane passengers is space time series data. Generalized Space Time Autoregressive (GSTAR) is an approach to forecast time series data that show a connection between time and location. The GSTAR model can be applied to data with heterogeneous location characteristics. In this research, the GSTAR model is used to form a forecasting model for the number of domestic flight passengers at three major airports in Indonesia Juanda Airport, Soekarno Hatta Airport, and Hasanuddin Airport, using optimal location weighting. The residual of the GSTAR model satisfies the white noise assumption. The location weights used are uniform weights, inverse distance weights, and cross-correlation normalization. The outcome model is constructed with a first-order differencing, an autoregressive order of 1, and spatial order of 1, resulting in the formation of the GSTAR (11)I(1) model. The most optimal model generated is the GSTAR (11)I(1) model utilizing inverse distance weights due to its smallest sMAPE value compared to other weightings, which stands at 8,71%. This sMAPE value signifies a highly precise forecast accuracy.]]>
Copyrights © 2026
