Citation

Belgaum, Mohammad Riyaz and Suresh, Kummari and Tharun, Meeniga and Sreenivasulu, Nandyala and Satyanarayana, C. V. (2025) Comparative Analysis of Detecting Anomalies in Real-Time Streaming Data. In: International Conference on Data Science and Applications, 17-19 July 2024, Jaipur, India.

Full text not available from this repository.

Abstract

The scheme is similar to the current state-of-the-art anomaly in pathfinding and is less demanding in terms of storage and time. One type of data analysis that looks for unusual events in the system is outlier detection, also known as outlier detection. Anomaly detection algorithms are checkpoints for traffic at every level, from the IoT network level to the data center. Outliers or anomaly detection can be checked using the Box-Whisker method or Density-Based Spatial Clustering of Application with Noise (DBSCAN). Besides clustering, DBSCAN can also be used for anomaly detection. One of the distinguishing features of DBSCAN is that it divides content into three categories: core content, fringe content, and popular content. For unrelated objects, the Euclidean distance method is used. Depending on the target, data points that differ from the mean are removed from the analysis or solution to avoid bias, called outlier detection. Detection of anomalies in data flow is an important problem in many practical applications because it provides some important information such as network security attacks, fraud, or other time usage. Different methods have been developed to identify anomalies: exclusion-based statistical analysis and integration. The authors focus on Isolation Forest (iForest), a method for detecting anomalies. iForest and other state-of-the-art anomaly detection methods, however, have low storage and difficult predictions. Implementation of various techniques like Random Forest, Decision Tree, Hybrid Adaboost, and iForest are used for comparison, which shows that iForest outruns remaining other techniques in terms of accuracy with 99.2%, precision with 99%, and recall with 99%.

Item Type:	Conference or Workshop Item (Paper)
Uncontrolled Keywords:	Anomalies, data analysis
Subjects:	Q Science > QA Mathematics > QA150-272.5 Algebra
Divisions:	Faculty of Computing and Informatics (FCI)
Depositing User:	Ms Suzilawati Abu Samah
Date Deposited:	29 Jul 2025 00:35
Last Modified:	31 Jul 2025 21:00
URII:	http://shdl.mmu.edu.my/id/eprint/14320

Downloads

Edit (login required)