Fairness in Algorithmic Decision Making
Permanent URI for this collection
1 - 4 of 4
ItemInstance-dependent cost-sensitive learning: do we really need it?( 2022-01-04)Traditionally, classification algorithms aim to minimize the number of errors. However, this approach can lead to sub-optimal results for the common case where the actual goal is to minimize the total cost of errors and not their number. To address this issue, a variety of cost-sensitive machine learning techniques has been suggested. Methods have been developed for dealing with both class- and instance-dependent costs. In this article, we ask whether we really need instance-dependent rather than class-dependent cost-sensitive learning? To this end, we compare the effects of training cost-sensitive classifiers with instance- and class-dependent costs in an extensive empirical evaluation using real-world data from a range of application areas. We find that using instance-dependent costs instead of class-dependent costs leads to improved performance for cost-sensitive performance measures, but worse performance for cost-insensitive metrics. These results confirm that instance-dependent methods are useful for many applications where the goal is to minimize costs.
ItemFair CRISP-DM: Embedding Fairness in Machine Learning (ML) Development Life Cycle( 2022-01-04)With rapid adoption of machine learning (ML) technologies, the organizations are constantly exploring for efficient processes to develop such technologies. Cross-industry standard process for data mining (CRISP-DM) provides an industry and technology independent model for organizing ML projects’ development. However, the model lacks fairness concerns related to ML technologies. To address this important theoretical and practical gap in the literature, we propose a new model – Fair CRISP-DM which categorizes and presents the relevant fairness challenges in each phase of project development. We contribute to the literature on ML development and fairness. Specifically, ML researchers and practitioners can adopt our model to check and mitigate fairness concerns in each phase of ML project development.
ItemAchieving MAX-MIN Fair Cross-efficiency scores in Data Envelopment Analysis( 2022-01-04)Algorithmic decision making is gaining popularity in today's business. The need for fast, accurate, and complex decisions forces decision-makers to take advantage of algorithms. However, algorithms can create unwanted bias or undesired consequences that can be averted. In this paper, we propose a MAX-MIN fair cross-efficiency data envelopment analysis (DEA) model that solves the problem of high variance cross-efficiency scores. The MAX-MIN cross-efficiency procedure is in accordance with John Rawls’s Theory of justice by allowing efficiency and cross-efficiency estimation such that the greatest benefit of the least-advantaged decision making unit is achieved. The proposed mathematical model is tested on a healthcare related dataset. The results suggest that the proposed method solves several issues of cross-efficiency scores. First, it enables full rankings by having the ability to discriminate between the efficiency scores of DMUs. Second, the variance of cross-efficiency scores is reduced, and finally, fairness is introduced through optimization of the minimal efficiency scores.