Comparação de técnicas de aprendizado de máquina para predição da disponibilidade de bicicletas no projeto Bicicletar Fortaleza

Abstract

The Bicicletar project consists of a shared bike program located in the city of Fortaleza, Ceará. The project nowadays has a web system and a mobile system, which has as main functionality the provision of information of the bike stations, such as the free bikes numbers, location, status and amount of vacancy. However, through the main functionalities of the applications, the users can’t get information in advance about the availability of bikes at specific stations and times, causing delays and inconvenience to users of the program. This article proposes a comparison of machine learning techniques to predict free bikes on stations of the Bicicletar project.