Attribute value reduction for rule property preservation in variable precision rough set model

Variable precision rough set model, as an important probabilistic approach to rough set theory, can deal with many practical problems which involve noise data and cannot be effectively handled by Pawlak's rough set model. Generally, rough set theory based knowledge reduction includes attribute reduction and attribute value reduction. Attribute reduction in variable precision rough set model has been attracted many researchers' attentions. However, attribute value reduction in variable precision rough set model was rarely discussed. In this paper, an approach to attribute value reduction in variable precision rough set model is presented, with which the redundant information in the given decision table can be effectively removed and the properties of the acquired rules, namely deterministic or probabilistic, can be preserved well.