Monotonic decision trees on rough set theory in machine learning approach

摘要

The traditional theory of Rough Sets states objects by their discrete attributes, and does not account for the ordering of its values. We have introduced the technique of monotone discernibility matrix as well as monotones (object) reduce. Also monotone discrete functions theory, which is developed earlier for representation and computation of decision rules are explained. Feature's values as well as decision values are ordinal in various decision making tasks. Moreover, a monotonic constraint states that the objects having better feature values must not be designated to worse decision class. These types of problems are referred as ordinal classification having monotonicity constraint. A number of learning algorithms have been designed for handling these kinds of tasks. In this paper, we surveyed about the Monotonic Decision Trees.