To visualize the relative positions of decision making units (DMUs), this paper proposes a technique which combines a modified data envelopment analysis and an extended principal component analysis. Referring to a DMU with other related units, the former analysis works for evaluating the performance of the DMU and for determining the improvement direction. Introducing a fuzzy membership function, the latter analysis works for positioning DMUs in a two-dimensional graph. The typical feature of the proposed technique is to clarify the positional relationship among efficient DMUs which were not clear by a simple combination of the original data envelopment analysis and correspondence analysis. This paper also demonstrates how the proposed technique works for Japanese prefecture data.