Skip to main content

(EST-4062) Using Case Based Reasoning to Improve Top-Down Estimate Accuracy

Presentation Icon
Level: Advanced
TCM Section(s):
7.3. Cost Estimating and Budgeting
7.4. Resource Planning
Venue: 2023 AACE International Conference & Expo

Abstract: Conventional wisdom assumes that the larger the database, the greater the accuracy. This does not necessarily apply in case-based reasoning models. The accuracy of top-down estimating tools using case-based reason can be improved by using a small sample of highly similar projects.This estimating tool’s case-based reasoning algorithm compares characteristics between the planned project and projects in the database to identify historical projects for high levels of similarity and then uses these projects in a Monte Carlo simulation to produce a range of acceptable values based upon statistical confidence levels. This tool was validated using three historical projects producing estimates that fell within one standard deviation of the actual values by limiting the total number sample projects to ones that were highly similar for use in the simulation model. Reducing the sample size reliably improved accuracy of the estimated values for each simulation.