关键词: 粗糙集, 离散化, 区间型数据, 相似度, 关联度

Abstract: The area of knowledge discovery and data mining are growing rapidly. A large number of methods are employed to discrete data, however, most of the existing discretion methods are applied in the case of attributes with real-value. In the practical application, the attribute value is interval number in many cases. Aiming at this problem, a new discretization algorithm applied to interval numbers is proposed. Similarity degree of interval number is used to describe the similar relation of two interval numbers. Threshold degree is defined to ensure discrete relationship between the data to implement algorithm. A new variable-associated degree is proposed through analysing action of similarity degree in the algorithm, and associated degree is used to improve algorithm. A group of data set is applied to testing the performance of the algorithm and the experiment result is compared with other discretization algorithms. The experiment result shows that the algorithm is effective.

Key words: rough sets, discretization, interval number, similarity degree, associated degree