As times develop, more and more vehicle users in big cities. But this was not well anticipated by the city government. So congestion is a common thing in big cities. There have been many solutions to overcome congestion, but it is still not effective because the application is still not perfect. But congestion can be avoided if the congestion can be predicted. Traffic prediction can be done with the SUMO application (Simulation Urban Mobility). The simulation results will be processed by python with the SVM multiclass method. The processed product will become a reference for predicting congestion. The best result from prediction is 100% accuracy in scenario with 90% train data and 10% of test data.