Traffic flow prediction by incorporating weather information in Naïve Bayes Classifier
F.I. Rahman1,*, A. Hasnat2, A.A. Lisa2
Naïve models are easy to use in real-world applications due to their uncomplicatedness.
Traffic flow is directly affected by weather condition.
About 25% mean absolute percentage error (R2=0.81) is found in naïve Bayes classifier.
It can be applied in traffic management.
It can provide logistic support to transportation agencies.
Traffic flow prediction is a fundamental element of the Advanced Traveler Information System (ATIS) and the Advanced Traffic Management System (ATMS). Such information may assist travelers in making better route choices and departure time decisions. Better management and traffic controlling system can be taken by transportation agencies to reduce congestion using traffic prediction. But accurate prediction is a challenging task for transportation engineers. Naïve models are the easiest process for traffic flow prediction. In this study, weather information or conditions are used in the naïve model for better traffic flow prediction. Naïve Bayes Classifier (NBC) is used here because of its easy application and low computation cost. In this study, five months of historical traffic flow data are trained with weather conditions. Then, considering weather conditions, the traffic flow of one month is predicted. One hour interval traffic flow prediction considering weather information in the NBC has a 25.019% mean absolute percentage error and the R-squared value is 0.81.
To cite this article: Rahman FI, Hasnat A, Lisa AA. Traffic flow prediction by incorporating weather information in Naïve Bayes Classifier. Journal of Advanced Civil Engineering Practice and Research 2019;8:10-16.