GSTAR MODELLING TO FORECAST MANY VEHICLE ENTERING PURBALEUNYI TOLL GATES WITH SOME WEIGHT MATRIX PERSPECTIVE

• Main Author: Tashya Noviana, Nur
• Format: Final Project
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/39155
• Institution: Institut Teknologi Bandung

Weight Matrix is one of the most important thing to use Generalized Space Time Autoregressive (GSTAR) modeling. But, usually, weight matrix built based on assumption or subjectivity of the researcher. Minimum Spanning Tree (MST) can be one of the alternative to build matrix weight based on our own data. In this final project, many vehicles entering purbaleunyi toll modeled by GSTAR with some weight matrix perspective. According to STACF-STPACF graph, derived some suitable models, such as : GSTAR(1;1) modeling, GSTAR(1;2) modeling, GSTAR(2;1,1) modeling. Therefore, from comparison between root mean square error, Akaike’s Information Criterion (AIC), and Bayesian Information Criterion(BIC), we conclude that GSTAR(1;1) with matrix weight based on radius distance is the best model to forecast many vehicles entering purbaleunyi toll.