Posted by

Correlation Matrices Revisited

Risk II Track



In a cost risk analysis, risk distributions are sometimes assigned at a low level and then combined, most commonly using Monte Carlo simulation. When using this technique, it is important to use a realistic correlation matrix. It is usually impractical to know all the pairwise correlations. This session will address some of the techniques for completing a credible correlation matrix at differing levels of input information and certainty. Additionally, the session will show examples and review the mathematics behind legitimate correlation matrices. Past sessions have discussed these topics and typically deal with correction techniques for expert-based correlation matrices. This session will be different since it deals less with correcting expert based correlation matrices and more with dealing with uncertainty in the correlations or incomplete correlation matrices. It will also address an extension of the problem to a parametric analysis problem.


Steven Ikeler
Army Capabilities Integration Center
Mr. Ikeler is an Operations Research Analyst for Army Capabilities Integration Center (ARCIC) assigned to the Maneuver Center of Excellence at Fort Benning, Georgia. He specializes in Cost-Benefits Analysis for Soldier and Mounted Requirements Divisions. Prior to his arrival at ARCIC, Mr. Ikeler was an Operations Research Analyst for the Office of the Deputy Assistant Secretary of the Army for Cost and Economics (ODASA-CE) where he specialized in Cost Analysis. Mr. Ikeler graduated with a Master of Science Degree in Industrial Engineering (Operations Research) from Purdue University in West Lafayette IN. As of June 13, 2013 he is an International Cost Estimating and Analysis Association (ICEAA) Certified Cost Estimator/ Analyst.