A weighted support vector machine method and its application

Faced with the fact that training samples belonging to a normal operation status are much more than the ones belonging to an abnormal operation status, we present the weighted support vector machine method. When the weights of the penalty parameters for different classes satisfy a relation equation, the undesirable effect caused by the unbalanced training class size is reduced, and classification accuracy of an abnormal operation status is improved. Simulated experiments for the data of Wisconsin diagnostic breast cancer (WDBC) show the effectiveness of the method.