Skip to main content

(CSC-2107) Using Historic Data to Improve Monte Carlo Prediction of Project Outcomes

Primary Author: W. Craig Boudreau, P.Eng. CCP
Audience Focus: Intermediate
Application Type: Practice
Venue: 2016 AACE International Annual Meeting, Toronto, ON, Canada

Abstract: This paper presents a model using historic data from past projects to predict the final costs for similar projects through the use of Monte Carlo simulation. This model uses job-to-date measurements for ten (10) key performance indicators (KPIs) combined with known historic progression of these KPI’s to estimate the probable range of project cost at completion. Three different Monte Carlo models are developed, which vary based on inclusion or exclusion of input documents such as construction schedules and indirect staffing plans. To understand the historic progression of the ten (10) KPIs used in the Monte Carlo simulation, eleven (11) projects in the Western Canadian heavy industrial construction sector are examined.

This approach shows promise using a limited sample size of projects. Future work is recommended to increase the sample size used to derive the Monte Carlo model as well as the number of projects used to test the model.