The application of discretization based on rough set and information entropy in TCM

Mutual information is approved to be an effective method to analyze the correlation of physicochemical parameters with the symptoms and the syndromes and has been widely used in many applications. But there are some difficulties and weaknesses, i.e. the assumption of continuous attributes according with normal distribution, which is a hidden problem in the calculation. In this paper, rough set and information entropy are used to discretize the physicochemical parameters of blood stasis syndrome under the constraint that the consistence of decision table is invariant. And the discrete data is used to analyze the correlation of the four subsyndromes of blood stasis syndrome with the physicochemical parameters. The result indicates that this algorithm not only preserves the original continuous attributes´ information, but also improves the reliability of result and the calculation speed of the algorithm consumedly.