Enhancing Data Mining Efficiency: Performance Analysis of Svdd-Oma Based Outlier Detection System

Abstract

Outlier detection in data mining is an essential element, especially in terms of huge data, where discrepancies can reveal important trends or risks. The research is the center on the construction and performance evaluation of an outlair detection system using SVDD-OMA (supporting vector data details with customized mapping algorithms). The main goal is to improve the accuracy and scalability of external identification in complex and high-dimensional datasets. The functioning employs a large data analytics framework, including SVDD for border-based classification and OMA for dimensional deficiency and pattern optimization. Experimental evaluation reveals increased identity rates in various datasets, less false positivity and efficient processing time. Conclusions suggest that Svdd -ome clearly crosses traditional models about accurate and strength. The results of the study are especially relevant to industries such as finance, cyber security, and healthcare, where early discrepancy identity is necessary. This study presents a scalable and adaptive approach to detect real -time discrepancy in the Big Data System.