MODEL ESTIMASI BANGKITAN PERGERAKAN MODA LAUT MENGGUNAKAN METODE REGRESI LINIER DAN RANDOM FOREST

Abstract

The need for people and goods movement across regions have been increasing susbtantially over the years. Sea transportations plays pivotal role in allowing this trans-national movement.The ferry line connecting Langsa, Indonesia and Penang, Malaysia had been established to help people and good across borders, albeit no longer operating. This study aims to estimate trip generation and attraction between these ports and their surrounding areas, and suggest whether this ferry line should reoperate. In doing so, random forest was utilised alongside linear regression model. Random forest has been increasingly popular among researchers in making estimates or preditions, whether calculating numerical or categorical data. To develop both models, historical trip and socio-economics data were employed. It was found that Gross Domestic Regional Product (GDRP) and population significantly affect trip generation and attraction. Both variables shows positive effect on trip generation and attraction: an increase in either should increase trip generation and attraction. Similar to linear regression model, random forest model show an excellent predictive capability. With a sophisticated model incorporating more variables, it is expected that random forest should well exceed linear regression predicitve capability. As population and GDRP are expected to continue to grow, it is suggested that stakeholders should reoperate this Langsa-Penang ferry line.