Performance Analysis of ANN & KNN Classifiers for Fatality Severity Prediction in Road Traffic Accidents

Nowadays, a significant part of goods and passengers are transported on suburban highways with mainly high speed vehicles. Hence, these highways are very prone to accidents with different injuries. Due to the high fatality or severe physical/mental injury rates caused by car crashes, analyzing these accident-prone areas and identifying the factors affecting their occurrences is crucial. The specific objective of the study was to compare two machine learning classification methods, ANN (Artificial Neural Network) and KNN (K-Nearest Neighbor) in building classification models for the fatality severity of ۲۳۵۵ fatal crash data records during ۲۰۰۷-۲۰۰۹ occurred in the roadways of ۸ states in the USA. The investigations confirmed that KNN had a better performance than ANN with a higher accuracy and kappa rate of ۷۲% and ۶۳%, respectively. Moreover, classified fatality severity levels of the crashes were provided for both algorithms to generate risk maps on the roads, so as to create potential accident risk spots. These conclusions drawn provide valuable information for more accurate evaluation of fatality risk, as an improvement towards safety performance.

Keywords:

Classification , Fatality severity , Artificial neural network , K-Nearest neighbor , Accuracy , Risk map