George Rigopoulos

NeXCLass: A DECISION SUPPORT SYSTEM FOR NON-ORDERED MULTICRITERIA CLASSIFICATION

This paper presents NexClass, a Decision Support System (DSS) which supports classification of alternatives into predefined non-ordered categories according to their performance on evaluation criteria and implements a novel classification algorithm based on multicriteria analysis and outranking relations. More detailed, assignment to classes is based on the concept of non-exclusivity, which defines at what degree an alternative can be included in a specific category. For each category, a threshold is defined by the decision maker, which indicates its limit with respect to the evaluation criteria. Alternatives are next evaluated according to the criteria and non-excluding degrees are calculated for each category. Finally, an alternative is assigned to the category for which non-excluding degree gets the lowest value. NexClass DSS implements the above classification algorithm, providing a user-friendly interface, which supports decision makers to formulate and solve classification problems. In addition to the methodology and the DSS, we present a real world application at a classification problem in banking environment. Our findings derived from evaluation experiments in the banking environment provide valid evidence that the proposed methodology and the DSS effectively support decision makers in classification decisions.

Fuzzy Multicriteria Assignment For Nominal Classification: Methodology And Application In Evaluation Of Greek Bank'S Electronic Payment Retailers

This paper presents a novel multicriteria procedure for nominal classification problems. Assignment of an action to a category is based on the comparison between action and categorys least typical representative, which is considered as the category inclusion threshold. Evaluation is executed considering performance on evaluation criteria and calculation of a fuzzy inclusion relation for every action, generalizing preference relations model with concordance and nondiscordance concepts as used in ELECTRE methods. An application to a classification problem in banking is also presented in order to demonstrate its applicability to similar problems.

Multi-Agent Modeling And Simulation Of Consumer Behaviour Towards Payment System Selection

Digital retail payments have evolved during the last years mostly due to technology innovations. New actors enter the payments market, increase competition and threaten banks’ dominance. Consumers’ behavior is critical within this context, since they are the ones that will finally adopt or reject a payment innovation. In this paper we propose a model of consumer behavior towards payment system selection based on multi-agent simulation. It is based on a multi-agent platform that models consumers as intelligent agents and formats their behavior according to rules extracted from past payments data, experts’ knowledge and marketing research. Contributions from social science and network economics also refine the agent’s behavior in order to simulate accurately the domain. This paper presents the overall model and an initial analysis of payments domain, defining major actors in the domain and their interpretation to agents.

Military Student Assignment Using NexClass Decision Support System

Higher education programs in military academies often require the formation of teams, either ad hoc or permanent and the assignment of students into them based on a number of predefined criteria. The assignment of each student into a specific team or group is a time consuming task when executed manually and the number of criteria is large. In this work we propose a model for the assignment based on the application of NexClass methodology and decision support system. The approach is based on the utilization of multicriteria analysis in order to assign students into a number of categories according to predefined criteria. The overall approach is introduced along with an illustrative example.