01127225

Component

Aiming at verifying the performance of alternative methodologies for efficiency analysis of transit systems we carried a comparative study using DEA – Data Envelopment Analysis, and PCA – Principal Component Analysis for composing an efficiency index. Both approaches were applied using a data base containing information on public transportation variables from different Brazilian cities. Results obtained were compared and discussed in view of the study objectives. Although indices based on DEA and PCA yielded similar results when applied to certain systems and different when applied to others, the following conclusions were considered valid: (i) the PCA-based aggregate efficiency index gave results that are more consistent with the concept of efficiency; (ii) DEA’s intrinsic flexibility with respect to weight assignment to the analyzed variables may lead to optimistic assessments of systems, i.e. certain variables may be unrealistically emphasized in the analysis; (iii) PCA prints objectivity to the analysis, assigning weights to variables based on their contribution to the overall data base variability. Finally, DEA implementation may be challenging when a large number of variables is considered; PCA, on the other hand, was originally conceived to handle large data sets.

01120148

09-2029

English

16p

2009

Transportation Research Board 88th Annual Meeting

DVD

Figures (1) ; References (25) ; Tables (9)

Cities; Data collection; Database management systems; Economic efficiency; Public transit; Sustainable development; Urban areas

Principal Component Analysis

Brazil

Operations and Traffic Management; Public Transportation; Society

Transportation Research Board Annual Meeting 2009 Paper #09-2029

TRIS, TRB

Jan 30 2009 6:20PM