Treatment Method after Discretization of Continuous Attributes Based on Attributes Importance and Samples Entropy

It has great significance to efficiently distinguish the type of the samples´ data in the decision table after the discretization for the course of machine learning and data mining afterwards. This paper puts forward an annotation method of distinguishing the data type based on attributes importance and the samples entropy, and processed the simulation test using part of the UCI database which was artificially modified, it turns out the method is able to efficiently identify the data type with high accuracy, low misidentification rate and low reject rate.