Abstract

SABOGAL CARDONA, Orlando Antonio; HINCAPIE ZEA, Juan David; SANTA CHAVEZ, Jhon Jairo  and  ESCOBAR, John Willmer. LINEAL REGRESSION MODELS FOR FORECASTING OF TRAVEL TIME IN BUS RAPID TRANSIT. Cienc. Ing. Neogranad. [online]. 2015, vol.25, n.1, pp.77-89. ISSN 0124-8170.

One of the most important aspects in the implementation of Priority Systems (TSP) in mass transportation is to know the arrival and travel time of the buses requesting priority. This paper estimates the arrival times of buses using linear regression techniques for a simulation model of the mass transportation system of Pereira, Colombia. In particular, special attention to the validity of the assumptions of linear models in different corridors under different conditions is given. The simulations were performed in the Transmodeler® software and the analysis of the models were done using R programming language. The results show that it is difficult to construct a valid linear regression model. In addition, the violations of the assumptions of normality, independence and homoscedasticity are frequent. The only situation in which it was possible to build valid models was in an area without signalized intersections and stop stations. However, evaluations made to the travel time, response variables, and residual analysis indicate that generalized linear models should be used.

Keywords : Bus Rapid Transit (BRT); Linear Models; Regression Analysis; simulation; Systems Priority (TSP).

        · abstract in Spanish     · text in Spanish     · Spanish ( pdf )