Probability of Flaw Detection for Quasi-Separated Data

• Activity Number: 581 - Advancement in Theoretical and Applied Aspects of Modeling
• Type: Contributed
• Date/Time: Wednesday, July 31, 2019 : 2:00 PM to 3:50 PM
• Sponsor: Government Statistics Section
• Author(s): Christine Henry* and Christine Schubert Kabban
• Companies: Air Force Institute of Technology and Air Force Institute of Technology
• Keywords: Flaw Detection; Logistic Regression; Resampling; Aircraft Safety; Bias-Adjusted Likelihood; Probability of Detection

Abstract

Probability of Detection (POD) studies using the hit/miss criteria often suffer from quasi-separation in the data, which leads to a lack of convergence for the maximum likelihood equation, making all results from the logistic regression questionable. To find a converging solution, three techniques showed the most promise: the Lasso, Firth's bias-adjusted likelihood technique, and bootstrapped ranked set sampling, a nonparametric resampling technique known to minimize separation. These methods were applied to data which suffers from quasi-separation, specifically, a probability of detection study for finding flaws in aircraft fastener heads using eddy current. The resulting parameter estimates and POD estimates will be compared and the usefulness of each method to overcome separation issues demonstrated. The overall goal and motivation for the use of this technique is to be able to accurately predict in post-processing the smallest flaw size an inspector could miss with the sensor even in the case of quasi-separated data.