Uncertainty Measures for Multi-Granulation Interval-Valued Decision Systems.

Uncertainty measures can effectively capture data features, and thus they serve as a fundamental tool to evaluate uncertainty in data analyses. In regard with the uncertainty measures, there are some studies in single grain interval-valued rough sets. As a novel uncertainty methodology, multigranulation interval-valued rough sets can deal with hierarchical issues in granulation processing, but they acquire less discussions on uncertainty measurement. Accordingly, this paper mainly constructs uncertainty measures regarding the multi-granulation interval-valued decision systems (MGIVDS). Firstly, MGIVDS are introduced. Secondly, some concepts of uncertainty measures are defined in multi-granulation interval-valued decision systems, including fusion θ-conditional entropy, fusion approximation roughness and fusion θ-decision entropy. Finally, properties of three types of information measures are obtained. Relevant researches offer novel insights into uncertainty measurement in multiple granulation and interval data.