A new group-based screening approach with visual presentation. (October 2021)

• Record Type: Journal Article
• Authors: Ma, Li-Ching
• Is Part Of: Computers & industrial engineering. Volume 160(2021)
• Publication Date: 2021-10
• Subjects: Multi-criteria decision-making -- Screening -- Visualization -- Group -- Mixed integer programming; Engineering -- Data processing -- Periodicals; Industrial engineering -- Periodicals; 620.00285
• DOI: 10.1016/j.cie.2021.107562
• Languages: English
• ISSNs: 0360-8352

Abstract

Highlights: Provide group-based screening solutions. Develop models to improve previous screening and displaying models. Reduce both the number of misclassifications and squared errors. Propose a visualization model to simultaneously allocate and group items. Help decision makers observe the decision context. Abstract: Screening is a process of filtering out items that are less likely to be selected, so that decision makers can more easily focus on evaluating items that more likely to be chosen in the smaller set. The case-based distance method is a popular approach for screening. However, previous case-based distance methods usually minimized the overall squared error or the number of misclassifications, possibly leading to an increased number of misclassifications or the overall squared error, respectively. In addition, the efficiency of individual screening may be insufficient, especially when there are a large number of alternatives. This study develops a new visual group-based screening approach. Revised models to improve previous screening and displaying models are also constructed herein. The concept of similarity upper approximation is then employed to provide group-based screening solutions. Lastly, a new visualization model is proposed to simultaneously allocate, group and screen alternatives. Compared with previous methods, this approach produces the lowest misclassification rate, while simultaneously yielding the smallest squared error. The individual alternatives, relationships among alternatives and cases, grouping relationships, and the acceptable ring can be directly observed through visual presentation. In addition, the proposed approach can provide flexibility in that a decision maker can choose to employ group-based or individual-based screening.