Analysis of Marine Corps Recruits in the Delayed Entry Program
Abstract
This thesis aims to assist the Marine Corps in solving the problems associated with achieving the objectives outlined in Talent Management 2030, a supplement to Force Design 2030. We analyzed 35 variables derived and originated from the Marine Corps Recruiting Command's applicant information system spanning fiscal year (FY) 15-22. This study utilized random forest models to identify the essential attributes contributing to the accurate classification of Marine Corps recruits that ship to basic training. We cross-validated nine separate random forest models against each FY testing data. The analysis revealed that our added variables significantly affected the models accuracy. Our added variables included the days in the Marine Corps delayed entry program, days between the date of enlistment and the initial strength test, and the applicant weight difference between the initial strength test day and ship day. Although our model demonstrated high specificity, it tended to make false positive errors when classifying applicants as shippers. When cross-analyzed, most models proved robust against each FY test data, with the exception of FY 22. Overall, our findings provide valuable insights into our added variables that aided in accurately classifying a Marine Corps recruit who ships to basic training and highlight areas for further investigation.
Document Details
- Document Type
- Technical Report
- Publication Date
- Sep 01, 2023
- Accession Number
- AD1224363
Entities
People
- Johnathan L. Thornton
Organizations
- Naval Postgraduate School