Estimate Validation and Bias Assessment: Ratio-to-Driver Method

摘要

Cost estimate validation is often mentioned in AACE? International literature but not described in depth. This paper describes the practice of cost estimate validation including a method called ratio-to-driver. Validation starts with the business establishing an objective in terms of a cost strategy that is captured in the basis of estimate document. The cost strategy defines the estimating approach in respect to desired base estimate bias (and every estimate is biased). Achievement of the cost strategy is the quality being assured by validation. Then, reliable, normalized metrics (cost estimating relationships in ratio form) are developed from a comparison set of projects drawn from an historical database (or obtained from some other reliable source). Database systems often do double-duty as validation tools; a precursor to the future of analytics and machine learning. The ratio-to-driver method applies the metrics in a logical, stepped sequence of comparisons that seeks to pinpoint the cause of variations. Because base estimate bias is a systemic risk, and validation measures bias, validation is also a first step in quantitative risk analysis. While a long-established practice, estimate validation is not defined in AACE? cost engineering terminology (RP 10S-90) and is only superficially covered in other estimating RPs. As such, this paper is intended as a basis for an RP that will be aligned with others that include validation or benchmarking. The primary effected RPs (with abbreviated titles) are: 31R-03 (estimate review), 34R-05 (basis of estimate), 35R-09 (estimate planning), draft CE-81 (estimate requirements) and 42R-08 (parametric risk analysis).