Predição de velocidades utilizando dados de sensores através de técnicas de aprendizado de máquina

Abstract

One of the growing problems in large urban centers is due to the delay in moving from one point to another in the city, due to traffic jams, peak hours, road works, etc. However, the current high availability of data can help to create solutions to this urban mobility problem. Because of this, this work proposes and evaluates Machine Learning approaches in order to identify the best technique for speed prediction from data obtained by a network of sensors spread over the city of Fortaleza (Brazil), during the year of 2014. For this purpose, several Machine Learning techniques and algorithms were tested, in order to identify the algorithm and its hyperparameters that best deal with the proposed problem.