Visualization of Cross-Efficiency Matrices Using Multidimensional Unfolding

Visualization of data envelopment analysis (DEA) problems is a relatively neglected topic in the DEA literature. This may be partly due to multidimensionality of the DEA problems, and partly because of under- estimation of the visualization usefulness. However, through information visualization, not only a vast amount of digits can be comprehensibly represented in a single map, but hidden patterns and structures of the data can be revealed through a bird's-eye view of the data object.Any DEA visualization method must choose a DEA dataset to be visualized, and a technique to graphically present the chosen data set. Such visualization technique usually is a dimensionality reduction method, since the DEA datasets are oftentimes multidimensional. This study suggests a method to visualize DEA cross-efficiency matrix (CEM), using multidimensional unfolding (MDU) technique.The suggested methodology is illustrated by means of two artificial datasets, and afterwards two real datasets have been visualized with the new method. The final maps can be used in anomaly detection, and since CEM is composed of the DMUs in both rating and rated aspects, the anomalies, such as maverick units or outliers, can be identified through a comprehensive approach. Nevertheless, the usage of this data exploration tool can go beyond anomaly detection, based on the goals of the researchers