Lessons Learned in Developing Cost Estimating Relationships

摘要

Typically, cost estimating models are probabilistic in nature as their objectives are to make statements about uncertain costs in the future. The uncertain item in the estimate can be viewed as a random variable and it is desirable to estimate certain aspects of the conditional distribution of the variable. A cost estimating relationship (CER) aims to describe the relationships between the dependent variable and cost variables or drivers while a correlation coefficient between cost variables measures their dependence. CERs are essential to develop a probabilistic cost estimating model. This paper will show how CERs can be developed and validated visually and interactively by using linear regression analysis techniques from cost data sets. In addition, lessons learned in analyzing cost data sets for cost estimating relations will be shared.