一种基于粗糙集属性频度约简算法的改进

为了获得有效的属性最小相对约简，在基于属性频度的启发式约简算法的基础上，提出了一种同时满足属性重要性和频度改进的启发式约简算法。该算法的基本思想是：以属性的核为基础，以频度作为选择属性的启发信息，即把属性频度最大的属性添加到核属性中，这样就把分类能力较强的属性添加到约简集合中，从而能够获得较优的约简。

An Improvement of Reduct Algorithm Based on Rough Set of Attributes Frequency

To obtain the minimal relative reducts of effective attributes,from the viewpoint of heuristic reduct algorithm based on attributes' frequency,propose a heuristic reduct algorithm,which satisfies both attributes' importance and amelioration of frequency.The main idea of algorithm is: the core of attributes is considered as the basis,the frequency is considered as the heuristic information for selecting attributes and then add the attributes with maximal frequency into the core attributes,from which the attributes with better ability for classification purpose can be joined the reducts,such reducts are preferable.