An Automated Machine Learning Approach for More Efficient Marine Corps Recruiter Prospecting

Abstract

The military recruiting environment is facing significant challenges, making recruitment goals more difficult to obtain. Due to these difficulties, the Marine Corps must find new ways to target the right demographics effectively. This thesis serves as a proof of concept for recruiting: can we employ automated machine learning to accurately prioritize public high schools using publicly available data? Current methods by Marine Corps Recruiting Command to prioritize high schools are largely unsystematic, potentially leading to inefficient allocation of recruiting resources. This study employs Microsoft Azure to demonstrate how we can use automated machine learning to enhance the efficiency of recruiting efforts. I find that automated machine learning using publicly available data may be an effective tool for predicting which public high schools to prioritize. Additionally, the automated machine learning predictions produced more contracts than the Marine Corps’ choices of priority schools. I recommend that the Marine Corps and other service branches further explore the use of automated machine learning and open-source data to enhance their recruitment strategies. Additionally, the key predictive variables identified by the automated machine learning model align closely with the criteria used by Recruiting Station leaders. However, the model provides a more granular analysis, enabling the identification of subtle patterns and interactions between each variable.