preprints.org > computer science and mathematics > artificial intelligence and machine learning > doi: 10.20944/preprints202302.0460.v1

Preprint Article Version 1 Preserved in Portico This version is not peer-reviewed

Version 1 : Received: 27 February 2023 / Approved: 27 February 2023 / Online: 27 February 2023 (08:42:14 CET)

The cost to train a basic qualified U.S. Navy fighter aircraft pilot is nearly $10M. The training includes primary, intermediate, and advanced stages, with the advanced stage involving extensive flight training and being very expensive as a result. Despite the screening tests in place and early-stage attrition taking place, 4.5% of aviators undergo attrition in this most expensive stage. Key reasons for aviator attrition include poor flight scores, voluntary withdrawals, and medical causes. Reduction in late-stage attrition offers several financial and operational benefits to the U.S. Navy. To that end, this research leverages feature extraction and machine-learning techniques on the very sparse flight test grades of student aviators to identify those with a high risk of attrition early in training. Using about 10 years of historical U.S. Navy pilot training data, trained models accurately predicted 50% of attrition with a 4% false positive rate. Such models could help the U.S. Navy save nearly $20M a year in attrition costs. In addition, machine-learning models were trained that can recommend a suitable training aircraft type for each student aviator. These capabilities could help better answer the need for pilots and reduce the time and cost to train them.

Pilot Training; Machine Learning; Attrition Prediction; Aircraft Suitability; Cost Savings; Automation

Computer Science and Mathematics, Artificial Intelligence and Machine Learning

* All users must log in before leaving a comment