|Friday, February 21|
|CS11 Risk Prediction & Modeling||
Fri, Feb 21, 1:30 PM - 3:00 PM
Risk Intelligent Modeling: Principles for Powerful Metric Based Risk Scoring (302752)Joeseph Michael, Deloitte and Touche LLP
*Robert J. Torongo, Deloitte and Touche LLP
Keywords: Analytics, Risk, Modeling, Metric Scoring
Metric based risk scoring can provide insightful and powerful ways to understand complex data relationships. It can be used to assess risk, understand performance, and to summarize performance measures that matter to an organization. If designed properly, metric scoring systems can give leaders and stakeholders an efficient and meaningful view of risk that impact their fiscal outlooks, operational decisions, and marketing strategies. This presentation will cover some important principles and practices for metric based risk scoring. In this presentation we discuss a variety of risk modeling approaches that we have applied in diverse data environments ranging from large and complex data systems to small tabular data sets. We will cover approaches such as: standardized risk metric analysis, cluster analysis based criteria modeling, regression based anomaly detection, risk identification using principal components analysis, methods for portioning small populations, and decision trees. Many of these approaches allow for intuitive and insightful customization when properly understood.