Title:
DESIGN OF TRAFFIC ACCIDENT PREDICTION MODEL IN TOLL ROAD USING A DECISION TREE ALGORITHM

Authors:
Wiwik Budiawan, Sriyanto, Singgih Saptadi, Ary Arvianto, Harun Pamuji and Pertiwi Andarani

Abstract:
A toll road is a road that the users are obligated to pay, which is held to improve efficient transportation services. Although toll roads have relatively more ideal conditions than highway roads, many traffic accidents still occur on the road. Toll road managers collect operational data on toll roads, including daily traffic, weather, and accident data. One of the solutions to increase the level of toll road safety is to design an accident prediction model through data mining. In this paper, the prediction model was made using attributes according to the framework consisting of day, type of road surface, weather conditions, road surface conditions, time of occurrence, driver sex, and type of vehicle. The prediction model was built to predict certain areas' probability and severity of accidents. The prediction model is built using the decision tree algorithm. The results show that the attributes used can predict the severity of accidents with 39.73% accuracy. The most vulnerable area is in section B on 9 to 10 km, with a total number of accidents of 13.17% of total accidents.

Keywords:
Traffic Accidents, Toll Road, Data Mining, Prediction Model, Decision Tree.

DOI:
http://dx.doi.org/10.52267/IJASER.2022.3602