EURO 2025 Leeds
Abstract Submission

1834. A new RDM-based EAT Model for efficiency analysis of non-homogeneous units with negative Data

Invited abstract in session MD-60: DEA methodological developments I, stream Data Envelopment Analysis and its applications.

Monday, 14:30-16:00
Room: Western LT

Authors (first author is the speaker)

1. Anjali Naik
Department of Mathematics, Indian Institute of Technology Delhi India
2. Ananya Sharma
Department of Mathematics, Indian Institute of Technology Delhi
3. Harshit Joshi
Department of Mathematics, Indian Institute of Technology Delhi
4. Aparna Mehra
Department of Mathematics, Indian Institute of Technology Delhi

Abstract

A new Data Envelopment Analysis (DEA) model is introduced in this paper to evaluate the efficiency of non-homogenous decision-making units (DMUs) with negative input-output data. We propose RDM-based convex DEA and Free Disposal Hull (FDH) models, specifically designed to handle negative data scenarios. The DEA and FDH models frequently exhibit overfitting; therefore, we employ the Efficiency Analysis Tree (EAT) method as an alternative. This method utilizes regression trees to identify production frontiers while adhering to microeconomic principles such as free disposability. The traditional EAT method caters to homogeneous DMUs, making it unsuitable for non-homogeneous scenarios. We suggest a new RDM-based EAT model for non-homogeneous DMUs for nonconvex technology and an RDM-based CEAT model for convex technology. The new RDM-based EAT and CEAT models for non-homogeneous units are compared with the RDM-based FDH and RDM-based convex DEA models for non-homogeneous DMUs, respectively. Additionally, we apply statistical tests to assess model differences, finding statistically significant variations among them. Furthermore, simulations conducted on randomly generated datasets examine performance trends across different conditions, including negative values.

Keywords

Status: accepted

Back to the list of papers