Exploring the Role of Graphical Exploratory Data Analysis in Understanding Statistical Patterns and Outliers

Posted: Oct 16, 2012

• Julian West
• Kaitlyn Gray
• Katherine Brooks

Abstract

This research presents a novel methodological framework that repositions graphical exploratory data analysis (GEDA) from its traditional role as a preliminary data inspection tool to a central analytical methodology for pattern recognition and outlier detection. While conventional statistical approaches often relegate visualization to supplementary status, our study demonstrates how systematic graphical exploration can reveal complex data relationships that remain obscured by purely numerical methods. We introduce the concept of 'visual inference chains'—sequential graphical procedures that enable analysts to trace the emergence of patterns and anomalies through progressive visualization layers. Our methodology integrates principles from cognitive psychology, information visualization, and statistical graphics to develop a comprehensive GEDA workflow that addresses the limitations of automated outlier detection algorithms. Through empirical evaluation across multiple datasets spanning environmental science, healthcare, and social media analytics, we demonstrate that our graphical approach identifies 37