Predicting Road Traffic Crash Severity in Kaduna Metropolis using some Selected Machine Learning Techniques

Abstract

Road Traffic Crash (RTC) is among the leading causes of death in the world and has a significant impact on the socio–economic development in a society. Generally, RTC can be caused by one or a combination of the following factors: Human, environment and vehicle.  This study utilized five data mining algorithm classifiers (Decision Tree (DT), K-Nearest Neighbor (KNN), J-Repeated Incremental Pruning to Produce Error Reduction (JRIP), Naïve Bayes (NB), and Multi-layer Perceptron (MLP)) to classify the severity of RTC and identify the significant causes of RTC in Kaduna State, Nigeria.  The RTC data used in this study included 26 RTC attributes with 1580 instances from 2016 to 2018 that covered fatal, serious and minor cases obtained from the Federal Road Safety Corps, Kaduna sector command. Two sets of experiments were performed on the classifiers (without and with feature selection). The study results showed that among the five data mining algorithms used, K-NN had the best accuracies of 94.8% and 96.1% respectively for the without and with feature selection experiments.