The consistency between the attribute relationships and time-serial count operators

The attribute construction with count operator can improve the accuracy of the prediction given by a data-mining model, but unrestricted use of this method can cause inconsistency problem of attribute relationships. After proposing three rules for the relationships between attributes and the inconsistency problem of attribute relationships, time-serial count operator, a new algorithm for time-serial correlative model, is proposed, which can avert the inconsistency problem of attribute relationships. Time-serial increment count operator can remarkably reduce the computing cost of time-serial count operator by a small precision cost when some assumptions are satisfied. The result of trials proves the above conclusion.