Expert Perception and Machine Learning Dimensional Risk Analysis

Our study develops a quantitative risk scoring system for Fidelity Investments' logical access data. We analyzed Snowflake warehouse data, focusing on associate, application, and account tables. Key risk attributes were identified through descriptive analysis. Expert surveys informed weighted risk score models, while Principal Component Analysis reduced dimensionality. Findings were integrated into an R Shiny dashboard for visualization and business recommendations. This system enhances understanding of risk sources and supports effective risk mitigation strategies. Future work will incorporate larger datasets and supervised machine learning techniques to refine the risk score framework's accuracy and applicability.