Empirical Study of a Reformulation of the Cumulative Average Learning Curve

摘要

One aspect of efficient management of resources that cannot be overstated is accurate cost estimation. The learning curve technique used in cost estimation continues to be a significant tool by itself and as an important factor in other cost estimation algorithms. This thesis conducts an empirical investigation of a theoretical reformulation of the cumulative average learning curve. The model is empirically corroborated by comparison of linear and nonlinear regression results with the classical unit and cumulative average learning curve specifications using two sets of aircraft production data. When autocorrelation was present and subsequently modeled into the data, the resulting linear models were significantly distorted whereas the non-linear models were not. While the model being scrutinized was adequate, the unit learning curve appeared to be the superior model.