Eine Plattform für die Wissenschaft: Bauingenieurwesen, Architektur und Urbanistik
Eine Plattform für die Wissenschaft: Bauingenieurwesen, Architektur und Urbanistik
Predicting Change Order Magnitude in Construction Projects—A Machine Learning Approach
https://orcid.org/http://orcid.org/0009-0009-8115-6229
https://orcid.org/http://orcid.org/0000-0002-2792-6781
https://orcid.org/http://orcid.org/0000-0003-1705-1093
Each construction project is unique with specific characteristics meaning that unavoidable uncertainties can happen during the project lifecycle. A well-defined change order management process right at the outset of the project enables the needed flexibility component that helps project stakeholders on dealing with inaccurate estimations, unforeseen conditions, design developments, among others. On the other hand, considering cost, time, and quality as three major project success factors, change orders can also disrupt their trade-off. Thus, it is crucial for the parties involved in construction projects to be able to identify change orders' performance expectations prior to engaging to a project. Additionally, for ongoing projects, it is essential to have a proper overview of potential change orders to reduce the negative effects before happening. This paper attempts to quantitatively analyzing the predictability of change orders’ magnitude in construction projects through machine learning techniques. To this aim, the following objectives and methods are followed: (i) predicting change orders’ magnitude as a three-class classification problem using SVM (Support Vector Machine), and Random Forest (RF) classifiers on three distinct datasets containing information on 2002 projects and more than 87,000 change orders, and (ii) investigating the best performing prediction model when being subjected to the identification of projects with different level of change orders. Results show that change order magnitude is predictable with around 75% accuracy in a 3-class prediction scenario, and it can be improved to 87% when targeting a specific class of the target attribute in binary classification scenarios. This research supports stakeholders involved in construction projects by pinpointing problematic projects as early as in the bidding stage while offering more realistic insights into the expected change orders in active projects.
Predicting Change Order Magnitude in Construction Projects—A Machine Learning Approach
https://orcid.org/http://orcid.org/0009-0009-8115-6229
https://orcid.org/http://orcid.org/0000-0002-2792-6781
https://orcid.org/http://orcid.org/0000-0003-1705-1093
Each construction project is unique with specific characteristics meaning that unavoidable uncertainties can happen during the project lifecycle. A well-defined change order management process right at the outset of the project enables the needed flexibility component that helps project stakeholders on dealing with inaccurate estimations, unforeseen conditions, design developments, among others. On the other hand, considering cost, time, and quality as three major project success factors, change orders can also disrupt their trade-off. Thus, it is crucial for the parties involved in construction projects to be able to identify change orders' performance expectations prior to engaging to a project. Additionally, for ongoing projects, it is essential to have a proper overview of potential change orders to reduce the negative effects before happening. This paper attempts to quantitatively analyzing the predictability of change orders’ magnitude in construction projects through machine learning techniques. To this aim, the following objectives and methods are followed: (i) predicting change orders’ magnitude as a three-class classification problem using SVM (Support Vector Machine), and Random Forest (RF) classifiers on three distinct datasets containing information on 2002 projects and more than 87,000 change orders, and (ii) investigating the best performing prediction model when being subjected to the identification of projects with different level of change orders. Results show that change order magnitude is predictable with around 75% accuracy in a 3-class prediction scenario, and it can be improved to 87% when targeting a specific class of the target attribute in binary classification scenarios. This research supports stakeholders involved in construction projects by pinpointing problematic projects as early as in the bidding stage while offering more realistic insights into the expected change orders in active projects.
Predicting Change Order Magnitude in Construction Projects—A Machine Learning Approach
https://orcid.org/http://orcid.org/0009-0009-8115-6229
https://orcid.org/http://orcid.org/0000-0002-2792-6781
https://orcid.org/http://orcid.org/0000-0003-1705-1093
Each construction project is unique with specific characteristics meaning that unavoidable uncertainties can happen during the project lifecycle. A well-defined change order management process right at the outset of the project enables the needed flexibility component that helps project stakeholders on dealing with inaccurate estimations, unforeseen conditions, design developments, among others. On the other hand, considering cost, time, and quality as three major project success factors, change orders can also disrupt their trade-off. Thus, it is crucial for the parties involved in construction projects to be able to identify change orders' performance expectations prior to engaging to a project. Additionally, for ongoing projects, it is essential to have a proper overview of potential change orders to reduce the negative effects before happening. This paper attempts to quantitatively analyzing the predictability of change orders’ magnitude in construction projects through machine learning techniques. To this aim, the following objectives and methods are followed: (i) predicting change orders’ magnitude as a three-class classification problem using SVM (Support Vector Machine), and Random Forest (RF) classifiers on three distinct datasets containing information on 2002 projects and more than 87,000 change orders, and (ii) investigating the best performing prediction model when being subjected to the identification of projects with different level of change orders. Results show that change order magnitude is predictable with around 75% accuracy in a 3-class prediction scenario, and it can be improved to 87% when targeting a specific class of the target attribute in binary classification scenarios. This research supports stakeholders involved in construction projects by pinpointing problematic projects as early as in the bidding stage while offering more realistic insights into the expected change orders in active projects.
Predicting Change Order Magnitude in Construction Projects—A Machine Learning Approach
Lecture Notes in Civil Engineering
Desjardins, Serge (Herausgeber:in) / Poitras, Gérard J. (Herausgeber:in) / Nik-Bakht, Mazdak (Herausgeber:in) / Nabipour, Nariman (Autor:in) / Martinez, Araham (Autor:in) / Nik-Bakht, Mazdak (Autor:in)
Canadian Society of Civil Engineering Annual Conference ; 2023 ; Moncton, NB, Canada
Proceedings of the Canadian Society for Civil Engineering Annual Conference 2023, Volume 4 ; Kapitel: 11 ; 137-150
18.09.2024
14 pages
Aufsatz/Kapitel (Buch)
Elektronische Ressource
Englisch
Assessing cyber risks in construction projects: A machine learning-centric approach
| DOAJ | 2024
| Emerald Group Publishing | 2021
Predicting environmental impacts of construction projects
| British Library Conference Proceedings | 1994