Review: Different approaches for Outlier Detection in Data Stream Mining

Review: Different approaches for Outlier Detection in Data Stream Mining

1st Prashant V. Chauhan , 2nd Vijay K. Vyas , 3rd Darshan P. Upadhyay

Abstract— Stream data differs from regular data in that it is continuously generated by many applications, posing unique processing issues such as enormous, limitless, and idea drift. For the researcher, data mining was one of the most exciting fields of study. In today's world, people are attempting to extract more information from data with order to aid in day-to-day forecasting. Outlier detection is utilized in a variety of applications, including fraud detection, intrusion detection, environmental monitoring, and medical diagnostics, so it's important to spot outliers in data streams. Outlier identification is done in a variety of ways. For outlier detection in data streams, some of them employ the K-Means technique, which aids in the formation of a similar group or cluster of data points. In any application, outlier detection is critical. In this research, we examined various outlier identification algorithms for stream data in depth and provided the results.

Keywords—outlier, outlier detection, STORM, k-means, k medoids