The last decades, Intelligent Transport Systems are in the spotlight. There is an increasing scientific interest in this field, due to the continuous rise in the number of vehicles causing a major problem in urban areas. A lot of research has been introduced to propose different ways to deal with the congestion, mainly by better predicting the traffic status and improving its management. Researchers suggest that both the data and the methodology being used have a significant impact on the results. The effectiveness of the estimations of the traffic conditions has been on the increase due to the evolution of technology. For example, mobile data in combination with machine learning techniques have led to more accurate estimations. This paper focuses on data processing techniques and in the use of neural networks, for predicting the traffic status based on the relation between traffic flows and speed. As a case study, the process is checked by using data of the city of Thessaloniki, Greece. Two types of test were performed, the first predicts the speed of eight sequentially quarters while the second estimates the speed four steps forward of the date time. The results of both tests provide accurate predictions.
