Enhanced Construction Cost Estimation of Highway Projects using Emerging Statistical and Machine Learning Techniques
Author(s)
Li, Mingshu
Advisor(s)
Editor(s)
Collections
Supplementary to:
Permanent Link
Abstract
Several state departments of transportation (state DOTs) have encountered significant challenges in accurately estimating costs for their highway projects, often resulting in discrepancies between the states’ DOT estimates (owner’s estimates) and contractors’ submitted bids. These inaccuracies can lead to cost overrun, scope change, schedule delay, postponement, and cancellation of transportation projects, which are problematic for both owner organizations and highway contractors. There is a critical need to enhance the quality of construction cost estimates to efficiently allocate public funds and increase confidence in engineer’s estimates. Addressing this need, the overarching objective of this research is to advance construction cost estimation for highway projects through the application of emerging statistical modeling and machine learning techniques, examining cost estimation at varying levels of granularity for a comprehensive analysis.
The study first adopts a temporal perspective at the monthly level, investigating risk factors that affect the accuracy of the owner’s estimate. This level of analysis allows for the examination of several variables representing the local highway construction market, overall construction market, macroeconomic conditions, and the energy market to identify leading indicators of the ratio of low bid to owner’s estimate. Appropriate time-series models, such as ARIMAX, will be applied to forecast this ratio using identified leading indicators. This macro-level analysis offers foundational insights into market trends and economic factors influencing cost estimations, setting the stage for more detailed investigations.
Transitioning to the project level, the research conducts survival analysis to assess the relationship between several potential drivers and the likelihood of bid over r times owner’s estimate. By innovatively applying concepts and methods from survival analysis to construction cost estimation, this part of the study explores the impact of project-specific, bidder-specific, and external market characteristics on estimation accuracy. This project-level analysis provides critical insights into the dynamics at play within individual projects, complementing the broader market perspective obtained from the temporal analysis.
Finally, at the most granular pay item level, forecasting models for early-phase cost estimation of lump sum pay items (Traffic Control and Grading Complete) are developed using text-mining and machine learning techniques. This approach involves retrieving project information available at the early stages of project development through text analysis and examining various machine learning algorithms with identified key predictive features to select the best-performing model. By focusing on specific pay items, this level of analysis directly addresses the practical needs of designers and cost estimators, offering precise tools for early cost estimation and further enriching the comprehensive understanding gained from the previous analyses.
This research contributes to the body of knowledge through: (1) developing appropriate multivariate time-series models (i.e., ARIMAX models) to predict the ratio of low bid to owner’s estimate; (2) creating a Cox proportional hazards model to explain and predict the likelihood of bid over r times owner’s estimates; (3) developing machine learning algorithms to accurately estimate prices of lump sum pay item at early stages of project development. It is anticipated that the research outcome would help cost estimating professionals in transportation agencies better understand the risk factors and potential drivers of the deviation between owner’s estimate and low bids, prepare more accurate cost estimates and develop appropriate risk management strategies for enhanced decision-making. Through its multi-level analysis, the study provides significant insights into project planning, budget allocation, and construction cost management, thereby underscoring the critical role of integrating machine learning and statistical modeling techniques in enhancing the accuracy and reliability of cost estimations for highway projects.
Sponsor
Date
2024-04-24
Extent
Resource Type
Text
Resource Subtype
Dissertation