University of Twente Student Theses
As of Friday, 8 August 2025, the current Student Theses repository is no longer available for thesis uploads. A new Student Theses repository will be available starting Friday, 15 August 2025.
Predicting cost overruns on utility projects with data-driven models
Potkamp, G.D. (2025) Predicting cost overruns on utility projects with data-driven models.
|
PDF
4MB |
| Abstract: | The utility industry faces growing challenges due to climate change, urbanization, and frequent cost overruns in underground utility infrastructure projects. Predicting cost overruns is difficult because of high project uncertainty and reliance on manual, experience-based estimates. This study addresses the need for more reliable and efficient cost overrun prediction by developing a data-driven decision model using binary classification machine learning techniques. Based on a case from Bam, a major Dutch utility contractor, the study followed the CRISP data mining process, identifying 46 relevant features from 888 project records. Multiple models were compared, including random forest, decision tree, and gradient boosting. Recursive feature elimination and hyperparameter optimization were applied to refine performance. The random forest model emerged as the best, achieving an accuracy of 0.6367 and AUC of 0.7174, indicating adequate predictive power. Stakeholders found the model useful for practical cost overrun assessment. The study contributes to literature by showing that binary classification models can distinguish between projects with and without cost overruns in the utility sector. Limitations include the use of overly specific features and reduced test performance compared to training, suggesting a need for more generalizable inputs and further exploration of overfitting or data leakage. |
| Item Type: | Essay (Master) |
| Clients: | Bam, Nieuwleusen, Nederland |
| Faculty: | ET: Engineering Technology |
| Subject: | 56 civil engineering |
| Programme: | Civil Engineering and Management MSc (60026) |
| Link to this item: | https://purl.utwente.nl/essays/106449 |
| Export this item as: | BibTeX EndNote HTML Citation Reference Manager |
Show download statistics for this publication
Show download statistics for this publicationRepository Staff Only: item control page