Prasetyo, Restu Wijang (2025) Analisis Risiko Konstruksi Menggunakan Markov Decision Process dan Reinforcement Learning di PT Berkah Development. Undergraduate thesis, UPN Veteran Jawa Timur.
![[img]](https://repository.upnjatim.ac.id/style/images/fileicons/text.png) |
Text
21012010098-cover.pdf Download (1MB) |
|
Text
21012010098-bab1.pdf Download (229kB) |
|
Text
21012010098-bab2.pdf Restricted to Repository staff only until 11 July 2027. Download (415kB) |
|
Text
21012010098-bab3.pdf Restricted to Repository staff only until 11 July 2027. Download (296kB) |
|
Text
21012010098-bab4.pdf Restricted to Repository staff only until 11 July 2027. Download (749kB) |
|
Text
21012010098-bab5.pdf Download (153kB) |
|
Text
21012010098-daftarpustaka.pdf Download (166kB) |
|
Text
21012010098-lampiran.pdf Restricted to Repository staff only Download (399kB) |
Abstract
Residential construction projects at PT Berkah Development are often hampered by various risks such as adverse weather, material delays, and labor shortages. Current risk management is still reactive and unsystematic, frequently causing projects to fall behind schedule. This research aims to address this problem by developing an intelligent system capable of recommending the best risk mitigation strategies. By leveraging the Markov Decision Process (MDP) and Reinforcement Learning (Q-Learning) approaches, this study seeks to create an adaptive, data-driven solution to significantly improve project time efficiency. To achieve this objective, this study collected data from various sources, including project documents and direct interviews with the on-site team. This data was then used to train a Q-Learning model in Google Colab. In this model, each project risk level (state) is analyzed to find the mitigation action that provides the greatest reward—in this case, the most significant reduction in project duration. To ensure the results are reliable, the model underwent rigorous testing to confirm its consistency, stability, and dependability. As a result, the model proved capable of providing highly effective and context-aware strategy recommendations. For instance, when the risk is at a critical level, the model recommends using protective canopies. Meanwhile, to address the risk of material delays, the strategy of switching vendors became the primary option. The implementation of these intelligent strategies successfully reduced the project duration from 180 days to just 143.1 days, in other words, achieving a time efficiency of 20.5%. This demonstrates that the Reinforcement Learning approach offers not only a theoretical solution but also a practical tool capable of making project management more efficient, faster, and cost-effective. Keywords: Construction Management, Risk Management, Markov Decision Process, Project Optimization, Reinforcement Learning, Q-Learning
| Item Type: | Thesis (Undergraduate) | ||||||||
|---|---|---|---|---|---|---|---|---|---|
| Contributors: |
|
||||||||
| Subjects: | H Social Sciences > H Social Sciences (General) H Social Sciences > HA Statistics H Social Sciences > HB Economic Theory H Social Sciences > HC Economics H Social Sciences > HD Industries. Land use. Labor |
||||||||
| Divisions: | Faculty of Economic > Departement of Management | ||||||||
| Depositing User: | Restu Wijang Prasetyo | ||||||||
| Date Deposited: | 17 Jul 2025 07:47 | ||||||||
| Last Modified: | 17 Jul 2025 07:47 | ||||||||
| URI: | https://repository.upnjatim.ac.id/id/eprint/39409 |
Actions (login required)
 |
View Item |