Applying Neural Networks to Air Force Personnel Analysis

The principal objective of this task involved evaluating artificial neural networks for application to personnel modeling by examining areas representative of many personnel models. The four areas chosen were airmen re- enlistment, the determinants of reenlistment, and the effects of policy levers; pilot training and more specifically the likelihood of candidates successfully completing Undergraduate Pilot Training; projection of aggregate time series personnel flow rates; and productive capacity of airmen as it relates to aptitude and experience. In addition, the productive capacity analysis was expanded into a working computer prototype allowing the user to examine the effect of changing aptitude/experience mixes on productive capacity. Performance was compared against traditional techniques such as regression analysis. Artificial neural networks, Learning vector quantization, Probabilistic networks, Back propagation, Personnel system modeling.