Data mining is the technique of discovering patterns among data to analyze patterns or decision making predictions. Anomaly detection is the technique of identifying occurrences that deviate immensely from the large amount of data samples. Advances in computing generates large amount of data from different sources, which is very difficult to apply machine learning techniques due to existence of anomalies in the data. Among data mining techniques, anomaly detection plays an important role. The identified rules or patterns from the data mining techniques can be utilized for scientific discovery, business decision making, or future prediction. Several algorithms has been proposed to solve problems in anomaly detection, usually these problems are solved using a distance metric, data mining techniques, statistical techniques etc. But existing algorithms doesn’t give optimal solution to detect anomaly objects in the heterogeneous datasets. This paper presents statistical control chart approach to solve anomaly detection problem in continuous datasets. Experimental results shows that proposed approach give better results on continuous datasets but doesn’t perform well in heterogeneous datasets.
展开▼
