Credit-based network management by weighted Fuzzy C-means

Nowadays network management is no longer limited to the routine maintenance of software and hardware. This paper introduces credit scoring into campus network management systems by weighted Fuzzy C-means clustering and Support Vector Machines (SVMs) predictor. The records of abnormal events on a large campus network over the past two years are obtained and sorted. Improved ReliefF analyzes the weights of five attributes of records before classified by weighted fuzzy C-means clustering into four classes. N-fold method is then employed to train a SVM classifier for prediction. The results indicate that the SVM predictor outperforms our previous system effectively. The classification results also consist with our goal of focused management on a small amount of users and set up reasonable starting scores according to the class the user belongs to.