Prediction of Traffic Congestion Based on Time Series Dataset Number of Vehicles Using Neural Network Algorithm Prasetyo Wibowo Yunanto (a,b*), Rahmat Gernowo (a,c), Oky Dwi Nurhayati (d)
a) Doctoral Program of Information System, Diponegoro University, Jalan Imam Bardjo, SH. No.5, Semarang 50241, Indonesia
*prasetyo.wy[at]unj.ac.id
b) Information Systems and Technology, Universitas Negeri Jakarta, Jalan Rawamangun Muka, Jakarta 13220, Indonesia
c) Department of Physics, Faculty of Science and Mathematics, Diponegoro University, Jalan Prof. Soedarto, SH, Semarang 50275, Indonesia
d) Computer Engineering Department, Engineering Faculty, Diponegoro University, Jalan Prof. Soedarto, SH, Semarang 50275, Indonesia
Abstract
The increase in the number of vehicles that is not proportional to the increase in road infrastructure results in high traffic congestion. Traffic congestion basically repeats itself, especially at certain hours due to high mobility at those hours, for example at the time of leaving and returning from work. Repeated traffic congestion can also occur by day for example at the beginning of the week or the end of the week. Based on this fact, traffic congestion can actually be predicted for certain roads at certain hours and days if the history data is known. In this study, a traffic congestion prediction model is proposed based on a traffic congestion dataset obtained within one week for 24 hours. The results show that the Neural Network algorithm has succeeded in predicting traffic congestion with a value of RMSE 13,188 in a 94 learning cycle.
