Optimizing Defense Acquisition Cycle Time Using Random Forest Models
Open Access DepositedEmerging and evolving global threats necessitate new military capabilities to be developed. Development speed is essential to maintain the technological advantage for the US Department of Defense in any potential engagement. However, new capability delivery has been significantly delayed from their original estimates. The DoD have been challenged to produce more accurate cycle time predictions and understand methods of reducing cycle time. This Praxis utilizes random forest regressions as a method of generating accurate cycle time predictions and understanding predictor importance. Random forest regressions can handle various input types, making them ideal for testing and evaluating candidate predictors. The random forest regressions are trained on the Government Accountability Office’s Weapon Systems Annual Assessment reports, the Director, Operation Test and Evaluation reports, and the Selected Acquisition Reports from 2015 to 2024. After feature selection and hyperparameter tuning, the model was able to produce an adjusted R2 of 83.4. Additionally, it identified that budgetary variables such as development and procurement costs are the most important for cycle time estimates. The timing of program initialization relative to election cycles was also identified as an important predictor. Finally, program behaviors like software development approach and reuse also have high importance.
- 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.