Accident Severity Estimation from Environmental Factors with the Proposed Dataset Separation Method

Abstract

Traffic accidents can be thought of as being the whole combination of events that usually occur due to reasons caused by humans, vehicles, roadway, and environment, which are the basic elements of transportation, and these events most likely result in injuries, death, and huge financial damage. In this study, by making use of a dataset, received from the New Zealand Transportation Agency, it is estimated whether or not traffic accidents occurring on roads end up with minor or serious damage. Predictions are made by exploiting machine learning algorithms. A new method has been proposed for separating the dataset and the classification success results made in this way gave the most successful scores by 89%. Based on the high performance rates, it is shown that such methods can be used to assist in preventing possible traffic accidents