The closest target-setting models based on data envelopment analysis (DEA) provide substantial contributions to benchmarking for requiring minimal effort to improve the performance of decision-making units (DMUs). However, these target-setting models may become impractical when several real factors, such as technology level, environmental influence, and governmental policy, are considered. In this study, we develop a novel DEA approach to target setting and benchmarking path selection by considering such real factors. A new bounded-change target-setting approach based on context-dependent DEA, in which the distance of evaluated unit to the best-practice frontier is minimized, is proposed. This approach sets benchmarks for inefficient DMUs in the short term and selects a realistic benchmarking path in the long term. Our approach also suggests a cross-level benchmarking path that few DEA-based benchmarking studies concern. This approach is applied to set sequential benchmarks and identify realistic benchmarking paths for inefficient DMUs via energy efficiency evaluation in the transport sector of China.

Freight bus is a new public transportation means for city logistics, and each freight bus can deliver and pick up goods at each customer/supplier location it passes. In this paper, we study the route planning problem of freight buses in an urban distribution system. Since each freight bus makes a tour visiting a set of pickup/delivery locations once at every given time interval in each day following a fixed route, the route planning problem can be considered a new variant of periodic vehicle routing problem with pickup and delivery. In order to solve the problem, a Mixed-Integer Linear Programming (MILP) model is formulated and an Adaptive Large Neighborhood Search (ALNS) algorithm is developed. The development of our algorithm takes into consideration specific characteristics of this problem, such as fixed route for each freight bus, possibly serving a demand in a later period but with a late service penalty, etc. The relevance of the mathematical model and the effectiveness of the proposed ALNS algorithm are proved by numerical experiments.