Generalized additive models for estimating motorcycle collisions on collector roads

In the recent years, the frequency of motorcycle collision in Indonesia, especially in Surabaya, is constantly increasing. This article will explain the use of the Generalized Additive Models (GAMs) to estimate motorcycle collision on collector roads in Surabaya. This study uses GAMs with Gaussian distribution and logarithmic link function, as well as application of software R in data processing. The case study takes place on urban roads in Surabaya, Indonesia. In this study, 69 roads of 120 collector roads in Surabaya are selected. The final model of this study indicates the relationship between the frequency of motorcycle collision on collector roads with explanatory variables which consists of traffic volume, road length, accessibility, road width, number of lanes, and traffic speed. Increasing values of explanatory variables in the prediction model lead to increased risk of accidents. These findings are expected to be considered in programs planned to reduce motorcycle collision on collector roads in Surabaya and other cities.