abnormal data detection and learning their behavior by abnormality and satisficing theory

Learning of abnormalities is a considerable challenge in data mining and knowledge discovery. Exceptional phenomena detect among huge records of the database which contains a large number of normal records and very few abnormal ones. This is important to promote confidence to a limited number of records for effective learning of abnormality. In this study, a new approach based on the abnormality theory and satisficing theory presented for confidence improvement of abnormal data detection and learning. First, the borders of abnormal and normal behavior clear using a combination approach based on abnormality theory then, satisfied solution extracted by means of satisficing theory. Modified RISE method as a bottom-up learning approach implemented to extract Normal and abnormal knowledge. The efficiency of the proposed model determined by using it, for abnormal stock selection from the Iran stock market. The superior of the proposed method results toward the results of applying decision tree and support vector machine is considerable. Accuracy of proposed method measure by g-means index. The results show the capability of proposed approach in abnormality detection and learning.

Keywords:

Data Mining , Abnormality theory , Satisficing theory , Bottom-up learning