DEVELOPMENT OF MIDBLOCK CROSSING SAFETY MODEL FOR UNDIVIDED ROAD IN BANDUNG

• Main Author: Lisya, Mutia
• Format: Theses
• Language: Indonesia
• Subjects: Teknik sipil
• Online Access: https://digilib.itb.ac.id/gdl/view/46055
• Institution: Institut Teknologi Bandung

Walking is a daily activity required by humans and most walking activities need for crossing the road. This thesis discusses about the development of pedestrian crossing safety model on the undivided road in Bandung. In Indonesia, pedestrians are one of the most vulnerable to accidents with a percentage 16% of road traffic accidents. Based on the 2017 Road Safety Annual Report (BIGRS, 2018), pedestrians are victims of fatalities and injuries with a percentage of 23% on the streets of Bandung. This overcomes the urgent need to address pedestrian-specific transportation problems in Bandung. The level of crossing safety created by building two model which is the Multiple Linear Regression (MLR) model and the ordered probit (OP) model. The model was built using three parameters consisting of control characteristics, road characteristics, and traffic characteristics. The location consists of 30 road sections with marked crossings. Each model has three types with the first type having traffic characteristics variables divided by vehicle type, the second type has a traffic characteristic variables that is combined for all types of vehicles and number of lanes variable is separated, and the third type has traffic characteristics variables that are combined for all types of vehicles but are split by number of lanes and number of lanes variable are separated. Of the six models that were built later after going through the validation process will be selected to assess the level of safety of crossing the road in the undivided road in the city of Bandung. The result of the model shows that the Multiple Linear Regression type 1 model becomes the best model with the value of Akaike Information Criterion (AIC) and Chi-square (X2) which are smaller than those for other models.