Dimensional Reduction of Data for Anomaly Detection and Speed Performance using PCA and DBSCAN
Anomaly detection is the major problem facing by many of industries. It includes network intrusion and medical sciences. Several fields like Astronomy and research also facing difficulties in finding effective anomaly detection. They have included several techniques to solve such problems. Clustering is the technique which has been employed by many of the researchers. The most commonly used algorithm to perform clustering is DBSCAN. It is well known clustering algorithm used in data mining and Machine learning. It is referred as Density based spatial clustering of application with noise. Because of its high complexity in computation, it must be decreased in terms of dimensionality of data points. PCA is a method used then to reduce dimensionality and produced a new data set which is again undergo DBSCAN. Here by the nature of the test results was precise there by such a methodology can be adjusted. The mix of PCA and DBSCAN was acutely confirmed and resultant examination shows that a speedup of 25% was improved while the quality was 80% diminishing the dimensionality of informational index of half.