A Defense Electronics Supply Chain Root Cause Analysis-based Model to identify data features that are viable for counterfeit predictive analytics
Open Access DepositedElectronic counterfeit parts are infectiously penetrating United States Department of Defense military systems, thereby threatening the lives of warfighters and national security (U.S. SASC, 2012; www-military-com, 2021; Moghadasi et al., 2022; U.S. DHS, 2023). Precedingly, the forerunner victims of counterfeit parts include the Defense Industrial Base as well as United States electronics suppliers who provide electronics parts to the Defense Industrial Base. Distinctively, although the United States Department of Defense is preemptively managing the risk of counterfeit parts through risk mitigation policies and procedures that generally advise contractors and subcontractors on how to detect and avoid counterfeit parts through process guidelines and utilization of probabilistic risk assessment, the specific causal factors or root causes of counterfeits are not always evident. As a result, contractors and subcontractors are inevitably driven towards a tactical implementation of United States Department of Defense policy instead of a strategic longer-term solution. This research praxis orchestrates a strategy that is a Root Cause Analysis-based model to identify data features that are viable for counterfeit predictive analytics. The proposed strategy includes traditional Root Cause Analysis steps to Collect Data, Decide How to Analyze Data, identify data features (causal factors) that are viable for predictive analytics as well as non-traditional steps such as building a predictive model, and a means to optimize the model. Subsequently, presented within this model are a list of highly viable data features that were identified following an evaluation of nine (9) statistical analyses: correlation, linear regression, graphical visualization, Naïve Bayes, univariate histogram analysis, the survival function, moving average, exponential smoothing, and logistic regression. The viable data features include number of counterfeiting incidents (state and company), U.S. semiconductor company inventories, U.S. semiconductor company number of DOD subcontracts, U.S. semiconductor raw materials, and U.S. semiconductor total debt (FBI, 2018-2022; ERA Inc, 2019-2022; www-usaspending-gov, n.d.; mergentonline-com, 2018-2022; U.S. DHS, 2023; www-bbb-org, n.d.; govtribe-com, 2024). Furthermore, two examples of predictive analysis were demonstrated within this praxis: logistic regression and Naïve Bayes whereby data features such as debt ratio, credit rating, defense relationship, DFARS compliance, and DOD Industrial Sector Membership were utilized to predict the probability that a company has fallen victim to counterfeiting (FBI, 2018-2022; ERA Inc, 2019-2022; www-usaspending-gov, n.d.; mergentonline.com, 2018-2022; U.S. DHS, 2023; www-bbb-org, n.d.; govtribe-com, 2024). Lastly, towards future direction, logistic regression and Naïve Bayes are areas for further evaluation and future research to assess quantitative risk of counterfeiting within U.S. Semiconductor companies or other organizations. Concurrently the objective of this strategy is to sustain a communication flow (feedback) of information exchange between the Defense Industrial base and the United States Department of Defense Acquisition offices.
- All rights reserved
Notice to Authors
If you are the author of this work and you have any questions about the information on this page, please use the Contact form to get in touch with us.