A Hybrid Model of Data Envelopment Analysis and Decision Tree for Efficiency Evaluation of Units (Case Study: Tehran Bus Transit)

Bus fleet performance measurement is under scrutiny of the urban policy makers because of the critical need to enhance the service level for the citizens. However، the situations that are involved in bus line efficiency are very important. Historical data of bus line performance are gathered daily by a wireless network in the city. In this study، bus lines’ data are investigated and processed by a hybrid method. Lines’ efficiency is measured by DEA technique، and then hidden rules in the data set are extracted in regard to the efficiency or inefficiency of the bus lines. At the end، the prediction ability of being efficient or inefficient has been evaluated by CART، C5. 0 and ensemble algorithms. The hybrid method has not been used in transportation field before; also the ensemble model that leads to higher accuracy has been used for the first time in this hybrid method. In this way، the combination of DEA and decision tree as two nonparametric tools will result in a road map to help policy makers for efficiency enhancement in urban transportation.