The US Air Force training enterprise is in the process of evolving to become more adaptive and responsive to the individual student. Initiatives such Air Education and Training Command’s (AETC’s) Pilot Training Transformation and those led by the Gaming Research Integration for Learning Laboratory (GRILL) have invested in technology and process improvements to deliver a more holistic, adaptive training environment to the warfighter. However, current data infrastructure has complicated modernization efforts. Existing training data from various sources are often stovepiped, use different data standards, and are difficult to cross-reference. Without a method to integrate data across sources, training and policy decisions cannot be driven by data. 

To address this issue, this paper discusses a modular framework for synthesizing training data across disparate sources to create actionable insights. The framework must accommodate data in the state it exists today, be adaptable to change, and inform the data infrastructure of tomorrow. This paper will discuss an exploratory effort to integrate student pilot data across multiple sources and data standards into meaningful insights that can drive training. Overall this integration across data sources will enable a more holistic view of student proficiencies to provide a basis for data-driven decisions on optimal training pathways tailored specifically to the student.

We illustrate this framework by integrating data from two primary sources (gradebook data and raw time series data logs from flight simulators) to better understand performance on specific maneuvers. First, relevant items and their associated grades are extracted from existing training gradebooks. Second, machine learning algorithms extract maneuver-level information from simulator data and quantify performance. Finally, data from both sources are used to predict performance on future maneuvers. Integration of disparate training data sources in this way offers a pathway to generate informed, data-driven decisions for Air Force training strategies.
 

Keywords
DATA;HUMAN PERFORMANCE;LEARNING ANALYTICS;TRAINING

Additional Keywords