Document Type

Conference Proceeding

Publication Date



U.S. Department of Defense (DoD) personal property moves account for 15% of all domestic and international moves - accurate prediction of their cost could draw attention to outlier shipments and improve budget planning. In this work 136,140 shipments between 13 personal property shipment hubs from April 2022 through March 2023 with a total cost of $1.6B were analyzed. Shipment cost was predicted using recursive feature elimination on linear regression and XGBoost algorithms, as well as through neural network hyperparameter sweeps. Modeling was repeated after removing 28 features related to shipment hub location and branch of service to examine their influence on algorithm performance. The best model resulted from a neural network hyperparameter sweep on the simple 4-feature dataset. That model contained 2 hidden layers of 250 neurons, possessed a mean absolute error (MAE) of $2162 on the holdout dataset, and had an overfitting measure of 0.2%. This model’s performance is a substantial improvement over a trivial mean-prediction model that possessed a MAE of $6856. Additionally, the model R2 = 0.87 compared favorably to existing work that achieved R2 = 0.73. Models that included 28 features related to shipment hub location and branch of service did not improve performance, showing that those features are insignificant in predicting shipment cost. All OLS regression and XGBoost models showed that the most influential features (in order) were weight, month of shipment, distance shipped, and days in transit.


The authors declare this is a work of the U.S. Government and is not subject to copyright protections in the United States.

[*] Author note: Tiffany Tucker was an AFIT DACS certificate student at the time of publication.

Source Publication

World Congress in Computer Science, Computer Engineering, and Applied Computing, Las Vegas, NV, 2023