Considering undesirable variable in PCA-DEA method: a case of road safety evaluation in Iran

This paper presents an approach composed from Data Envelopment Analysis (DEA) model and a multivariate statistical method, Principle Component Analysis (PCA), considering undesirable input and output variables. PCA is used to improve discrimination power of DEA and making variables as independent as possible to avoid overlapping of Decision Making Units (DMU’s) information. The advantage of the proposed approach is considering undesirable output and input variables simultaneously in PCA-DEA method; furthermore it was applied for performance assessment of different province’s road safety level in Iran.