by Anthony DeMarco
| September 19, 2014
The following is an extract from a paper written in 1978 from one of the founders of PRICE Systems:
Two questions are often asked by those unfamiliar with TruePlanning’s approach to cost modeling: What is your CER (cost estimating relationship)? And what is your data base?
These questions are closely related. Both are based on the assumption that the PRICE modeling approach is the same as that customarily used in developing cost estimating relationships. This is not the case.
The customary approach is to first gather as much relevant data as possible, then screen the data for consistency, reduce the data by formal statistical procedures and present the results in the form of one or more estimating relationships.
In contrast, the PRICE approach is process oriented rather than data base oriented.
Models are designed to emulate the processes by which experienced managers, engineers and cost estimators assess the impacts of key cost and schedule drivers.
The strength of the classical approach is that it enables investigators to test hypotheses and identify significant factors that have affected past developments. PRICE does not ignore these results. Indeed, similar procedures are often used in preliminary model development. As a consequence, there is nothing in the PRICE models that is inconsistent with classical results. The difference is that these results are not an end in themselves. When blended with quantifications of other subjective, but no less valid, perceptions by experience people, they contribute substantially to the PRICE methodology.
The weakness of the classical approach is that it is data base limited. It does not extrapolate well when applied as a cost estimation procedure to new situations. In fact, it usually does not even adequately describe the underlying data base. Data definitions are often inconsistent, and data poorly recorded that fitted equations are questionable at best. Extensive screening to eliminate inconsistent and unrepresentative data results in so few cases that the richness of single equation models is severely limited. These models simply cannot account for all of the factors that drive costs.
The principal reason for the success of the PRICE model is that it does not depend on a single CER or on a single data base. By focusing on the process of rational cost estimation, it preserves for the user the flexibility to tune the model to the particular data base most relevant to the estimate in question. This will normally be cost histories of previous projects within the users’ own organization or product line.
In effect, the PRICE model develops a new family of CERs to fit each specific application based on the data base it is tuned against.