Exploring the Application of Factor Analysis in Reducing Data Dimensionality and Identifying Underlying Constructs

Posted: Jan 09, 2023

• Alexander Thompson
• Daniel Moore
• Elijah Allen

Abstract

The exponential growth of data collection capabilities across scientific and commercial domains has created unprecedented challenges in managing and interpreting high-dimensional datasets. Traditional dimensionality reduction techniques, while effective in preserving variance, often sacrifice interpretability, leaving researchers with transformed variables that lack clear conceptual meaning. Factor analysis, a statistical method with roots in psychology and social sciences, offers a promising alternative by focusing on identifying latent constructs that explain the covariance among observed variables. However, its application in modern computational contexts remains underexplored, particularly in domains beyond its traditional purview. This research addresses the critical gap between variance-focused dimensionality reduction and interpretable latent structure discovery.