Short-term traffic congestion forecasting using hybrid metaheuristics and rule-based methods: A comparative study Pedro Lopez-Garcia 1 , Eneko Osaba 1 , Enrique Onieva 1,2 , Antonio D. Masegosa 1,2,3 , and Asier Perallos 1,2 1 DeustoTech-Fundacin Deusto, Deusto Foundation, 48007, Bilbao, Spain. 2 Faculty of Engineering, University of Deusto, 48007, Bilbao, Spain. 3 IKERBASQUE, Basque Foundation for Science, 48011, Bilbao, Spain. e-mail: {p.lopez, e.osaba, enrique.onieva, ad.masegosa, perallos}@deusto.es Abstract. In this paper, a comparative study between a hybrid tech- nique that combines a Genetic Algorithm with a Cross Entropy method to optimize Fuzzy Rule-Based Systems, and literature techniques is pre- sented. These techniques are applied to traffic congestion datasets in order to determine their performance in this area. Different types of datasets have been chosen. The used time horizons are 5, 15 and 30 minutes. Results show that the hybrid technique improves those results obtained by the techniques of the state of the art. In this way, the per- formed experimentation shows the competitiveness of the proposal in this area of application. Keywords: genetic algorithms, cross entropy, classification, machine learning, hybrid optimization, fuzzy rule-based systems, intelligent transportation systems 1 Introduction According to the Eurobarometer [2], road congestions are one of the problems that citizenship are more worried about regarding road transport. Therefore, traffic congestion prediction is a fundamental issue in the field of Intelligent Transportation Systems (ITSs). If congestion is predicted successfully, it could help to take decisions can result in noise reduction and energy savings. Also, it could increase the effectiveness and the performance of transport systems, and lead to savings in public infrastructure. While two of the most frequently used methods for this task in the last decade are the Kalman Filter and the Autoregressive Integrated Moving Aver- age (ARIMA), other alternatives have been developed in recent years. Among them, Soft Computing techniques as Support Vector Machines (SVM), Neural Networks (NN), Genetic Algorithms (GA) or Fuzzy Rule-Based Systems (FRBS) have been used in traffic forecasting tasks in particular [22], and in ITS field in general [17, 18].