An ensemble regression approach for bus trip time prediction

• Author(s): João M. Moreira ; Jorge Freire de Sousa ; Alípio M. Jorge ; Carlos Soares
• Date: 2006
• Persistent ID: http://hdl.handle.net/10216/6752
• Origin: Repositório Aberto da Universidade do Porto
• Subject(s): Ciências tecnológicas; Engenharia
• Document Type: Conference Object
• Language: Portuguese

Description

This paper is about bus trip time prediction in mass transit companies. We describe the motivations to accomplish this task and how it can support operational management on such companies. Then, we describe a Data Mining framework that recommends the expected best regression algorithm(s), from an ensemble, to predict the duration of a given trip. We present results that show the advantage of using an ensemble regression approach.