Consistency and Inconsistency in the Case of a Stochastic Paired Comparison Model

• Published in: 2024 IEEE 3rd Conference on Information Technology and Data Science (CITDS) pp. 1 - 6
• Main Authors: Mihalyko, Csaba, Orban-Mihalyko, Eva, Gyarmati, Laszlo
• Format: Conference Proceeding
• Language: English
• Published: IEEE 26.08.2024
• Subjects: 3 options; Computational modeling; consistency; Data models; Data science; inconsistency measure; Information technology; Modified Bradley-Terry model; Phase change materials; Production; Reliability; Stochastic processes
• DOI: 10.1109/CITDS62610.2024.10791369

The concept of consistency is a central question if pairwise comparison matrices based methods are applied but it is less investigated in such models which have stochastic background. Nowadays incomplete comparisons are in the focus of the research, namely how the traditional measurements can be generalized for such cases when some pairs are not compared. Data consistency in case of stochastic models has been previously investigated if two options (better and worse) are allowed in decision. In this paper we focus on the concept of consistency in the case of three options in a model defined by Davidson satisfying Luce's axiom. We present a definition for consistency and a measurement of inconsistency. From practical point of view, it can characterize the comparison results and can provide possibility for the decision makers to decide whether data are suitable or not to make reliable decision based on them.