Clustering and Perdiction of road Accidents

Due to urbanization and wide use of technology in life, road traffic accidents annually cause 1.2 million deaths and injury of more than 50 million people over the world. Thus, one important traffic issues, is prediction of this road traffic accidents and presentation the solutions to reduce them.

Clustering is one of the most widely used models for the prediction and identification of patterns that is used in this research. For this purpose, road traffic accident data of Fars province accidents is considered. After pre-processing, data is clustered using self-organizing neural network algorithm. Then the accident are classified into three classes such as low, medium and high and accident patterns are recognized using decision tree model.

In this research the road traffic accidents in Fars province were clustered into 11 clusters. Then the accident were classified into three classes such as low, medium and high and accident patterns were recognized using decision tree model. Finally some suggestions are offered to reduce road accidents.