Time-series method for predicting human traffic flow: A case study of Cañar, Ecuador

The forecasts of the flow of people in urban public transport units can help governments and townships to making major decisions for efficient management of their cities, to improve their public transport service infrastructure and providing a better quality of life for their community. In this paper, we present a system that uses a traffic flow prediction model, based on neural networks, for real-time analysis of users (people) that entering to transport unit (bus) at specific stops by tagging their mobile devices. In the prediction model we perform a time correlation analysis on the data collected from the flow of people that is obtained using a WLAN network within an urban transport unit. An LSTM network model was used, this model generates an adequate performance when forecasting the number of users that could be transported. Finally, the results of this system are displayed, analyzed and stored in a WEB and SQL server, the experimental results show that our system is a solid alternative when it comes to forecasting and monitoring crowds of people in real time in transport systems.