SVM based multi-index evaluation for bus arrival time prediction

The real time prediction of bus arrival time is important in public transport service. Many researches use a traditional error metrics to measure the predicting accuracy, which cannot evaluate the prediction service comprehensively. This paper proposes a novel multi-index evaluation method based on Support Vector Machine (SVM) for bus arrival time prediction. This method uses three new indexes including GPS coverage, release rate, and accuracy rate to evaluate the prediction service, and then uses SVM to train the model for multi-index evaluation. Experiment results show that our method is intuitive and comprehensive by using SVM based multi-index evaluation, and can position issue accurately according to the three new indexes.