Study on Anomaly Detection Algorithm of QAR Data Based on Attribute Support of Rough Set
Yang Hui, Chunjing Xiao, Yongwei Qiao · International Journal of Hybrid Information Technology · 2015
According to the characteristics of the large amount of QAR data，such as many parameters, time constraints, strong randomness and the problems of discrete data, together with attribute reduction and rules colleting during QAR anomaly detection, the paper proposed a anomaly detection algorithm of QAR data based on attribute support of rough set. Firstly, we discrete QAR data after converting the time sequence data into non-time sequence information system based on attribute support. Second, we carry on the attribute reduction using difference matrix based on statistics. At last, we get the fault rules with the decision tables based on fault occurrence statistics of the decision attributes. The method not only primely retains the time characteristics of QAR data, but also strengthens the relation between condition attributes and decision attributes. The efficiency of attribute reduction and rules extraction is high and the amount of calculation is small. The experimental results show that the method is feasible, reliable and availability.