ART - Predicting Engineering and Schedule Procurement Cost Growth for Major DOD Programs 2007

The document discusses the development of a model to predict the expected percentage of cost growth for a program, specifically in the engineering cost category and the schedule cost category. The document describes the methodology used, which is based on logistic regression models. Two OLS regression models are built, one for each cost category. The document includes figures showing histograms and quantile plots of the original cost variance data. The OLS models' mathematical structure is explained, and the parameter estimates and p-values for the explanatory variables are provided in tables. The document also discusses the caution needed when interpreting the variables and emphasizes that the models show mathematical associations, not cause and effect relationships. The document concludes by stating that the models passed the statistical tests for normality, constant variance, and independence, and no multicollinearity was detected. Cook's Distance analysis was used to identify influential data points, and one data point had to be removed from each model. Additionally, it is mentioned that for the Engineering Cost Variance % model, the program used as an example was the Cheyenne program.