Detecting Outliers in High Dimensional Data Sets using Z-Score Methodology

Abstract

Outlier detection is an interesting research area in machine learning. With the recently emergent tools and varied applications, the attention of outlier recognition is growing significantly. Recently, a significant number of outlier detection approaches have been observed and effectively applied in a wide range of fields, comprising medical health, credit card fraud and intrusion detection. They can be utilized for conservative data analysis. However, Outlier recognition aims to discover sequence in data that do not conform to estimated performance. In this paper, we presented a statistical approach called Z-score method for outlier recognition in high-dimensional data. Z-scores is a novel method for deciding distant data based on data positions on charts. The projected method is computationally fast and robust to outliers’ recognition. A comparative Analysis with extant methods is implemented with high dimensional datasets. Exploratory outcomes determines an enhanced accomplishment, efficiency and effectiveness of our projected methods.