Stochastic Conceptual Cost Estimating of Highway Projects to Communicate Uncertainty Using Bootstrap Sampling

Abstract

Conceptual cost estimating is typically completed early in the project lifecycle when little design work has been completed. Because little information is known at this early stage, the estimate usually deviates substantially from the actual construction cost. When expressed as a deterministic value, an estimate often leads to a false inference of accuracy by those not familiar with the vagaries of conceptual cost estimating, making it difficult for an agency to explain cost growth. A stochastic conceptual estimate allows an agency to produce a probability distribution of the likely construction costs and address the level of confidence in a given estimate. Named probability distributions are readily available for developing a stochastic estimate in a great deal of commercial software. However, instead of fitting available distributions, this research generates an empirical distribution to express a cost estimate range. Creating empirical distributions eliminates assumptions required for selecting named distributions. The stochastic data-driven model developed in this paper combines artificial neural networks and bootstrap sampling in a case study of 189 highway projects to train and test the estimating model.

Cite this work

No public copy is stored for this record, so only the metadata is cited.

Stored only when both papers are in the corpus — 26 citations exist upstream.

Papers that cite the same references — the relational fallback while embeddings are still backfilling.