Building an Information Systems Model for Traffic Congestion in Muscat City

Abstract

Traffic accidents are the main cause of death, and resulting in complex traffic congestion.  They occur for many reasons, among which there are both technological and human factors. The accident can occur due to the fault of a tired driver, due to condition of the road surface or malfunction of the vehicle system. This work intends to build a model to predict the time hour/day/date of an accident on specific city and street location, which causes traffic congestion. The proposed model could alert the driver of an accident occurrence in specific location and time. Therefore, this strategy will minimize the traffic congestion as most of the drivers will change their path route. A large data set has been collected from the Traffic Database for the past 10 years from Royal Oman Police (ROP). Data Mining Methodology has been customized to build the forecasting model. 4011 Time series instances have been used to train the forecasting model. The MAE (Mean Absolute Error) and RMSE (Root Mean Square Error) have been used for Evaluation on training data. The MAE scored ~5.1, and RMSE scored ~6.8. The obtained results are promising and could be useful for improving road safety and traffic congestion strategy.