Towards Gaining Robustness in Inverse Data Envelopment Analysis Models

The inverse data envelopment analysis (IDEA) approach aims to estimate input and/or output levels when preserving efficiency scores. The input-oriented IDEA version seeks the input level for producing expected output and the output-oriented one estimates the output level under a given input level while efficiency remains unchanged. However, in many real-world applications, full and precise information may not be available to guarantee successful IDEA implementation. This study presents a novel approach to combat inherent uncertainty, resultantly, enabling a move towards robustness of IDEA models. We particularly focus on two cases in this research. The first case occurs where the amount of extra input is not certain and almost impossible to be precisely determined due to restricted budget, market volatility, political issues and other external factors. The second case is observed in situations where input and/or output data encompasses uncertainty that might be resulting from errors in data measurement, data clearing, vagueness in variables (e.g., customer satisfaction or quality) or other internal factors from organizations.