AETC partners with UTSA to refine pilot candidate selection

  • Published
  • By Dan Hawkins
  • Air Education and Training Command Public Affairs
More effectively targeting pilot candidates may help the Air Force solve its pilot manning shortage, and collaborating with the education sector to do so has proven beneficial for both partners.

In an effort to gain more insight into what attributes could best predict the success of pilot training candidates, Air Education and Training Command turned to the data analytics program at the University of Texas-San Antonio to help them refine and validate the Pilot Candidate Selection Method as part of an Educational Partnership Agreement.

“We use the PCSM right now to help us determine who gets an undergraduate pilot training slot,” said Lt. Col. Steven Dillenburger, AETC’s Studies and Analysis Squadron commander. “The intent in working with UTSA in this partnership was to help us gather the data to validate if the background data the PCSM tool indicates is a valid measuring stick to go by or perhaps needs updating.”

To get things started, AETC developed a problem statement and gathered raw data to provide to the UTSA team so they could begin building models that better predict a candidate’s PSCM score, their graduation or attrition rate and class ranking indictors for excellent performing students, Dillenburger said.

“This is a problem where we have a pretty good data set to begin with,” Dillenburger said. “We, with help from the Air Force Personnel Center, gathered a great deal of information on prospective pilot candidates before they were selected for undergraduate pilot training. We combined that pre-selection data with their performance data from their time in undergraduate pilot training to build the data sets.”

Starting in June 2019 and using approximately 10,000 sets of individual raw data, including both background data like students’ Air Force Officer Qualifying Test or college GPA, and their previous pilot experience, such as flying hours or instrument ratings, the data analytics team from UTSA input the data into an artificial neural network and analyzed the variations.

“UTSA really helped us to cleanse that raw data and make it work in their algorithm,” Dillenburger said. “What the study found was in line with what we value already from the PCSM, in that the AFOQT scores, number of previous flight hours and any potential previous aeronautical ratings most positively relate to a successful student.”

With studies like those on anthropometrics (height standards) and others, the Air Force continues to look at every aspect of the pilot candidate selection process.

“This study was really valuable because we are looking hard right now at rated diversity and the PCSM really values flight hours, which can be difficult to obtain for potential candidates depending on their socio-economic background,” Dillenburger said. “We want to make sure we aren’t eliminating potential candidates based on (an inconsequential) piece of data.”

Another interesting takeaway from the UTSA study was the finding that a student’s GPA in college didn’t have a positive correlation to whether a student graduated from pilot training or not, Dillenburger said, while noting many factors can affect that particular finding.

The results from the study, which was completed in late October 2019, will be briefed to representatives from the rated diversity initiative working group who are looking to identify and mitigate barriers to entry into rated careers. The data will help determine actions items moving forward, Dillenburger said.

The partnership with UTSA was beneficial, not just in terms of the data that was collected, but in growing the Air Force’s ability to excel in the data science field.

“UTSA really has grown a world-class, state-of-the-art, data analytics program, and we’re thankful for their help,” Dillenburger said. “Getting the chance to see what software and processes they are using has been extremely helpful to our Air Force data analytics specialists as we continue to grow and evolve our own organic capability.”

For UTSA, centered in the heart of “Military City, U.S.A.” and surrounded by the military, the chance to help find solutions that could affect the Air Force’s ability to execute their role in the National Defense Strategy was one way to build bonds.

“UTSA strongly values the relationship between the Air Force and the university, and this project is one of the ways that we can strengthen that relationship,” said Dr. Max Kilger from the UTSA data analytics program. “Developing a collaborative approach to solving issues that relate to our national security benefits our country and at the same time provides an opportunity for faculty and students to work collaboratively with the Air Force.”

Additionally, the opportunity to work with the Air Force was helpful in terms of the educational experience it provided to UTSA students.

“Real-world problems are an important part of the training for our graduate students,” said Dr. Paul Rad, UTSA data analytics professor and one of the leads for the project. “They get exposure to real-world data, which is often challenging and complex, and under faculty supervision apply what they have learned to solve interesting and important challenges.”