Muhammad Ardy Cahya, . (2026) PERANCANGAN SISTEM PREDIKSI WORK-IN-PROGRESS MANUFAKTUR BAJA MENGGUNAKAN PENDEKATAN ARSITEKTUR DECOUPLED MACHINE LEARNING. Skripsi thesis, Universitas Pembangunan Nasional Veteran Jakarta.
![[img]](http://repository.upnvj.ac.id/style/images/fileicons/text.png) |
Text
ABSTRAK.pdf Download (140kB) |
|
Text
AWAL.pdf Download (714kB) |
|
Text
BAB I.pdf Restricted to Repository UPNVJ Only Download (196kB) |
|
Text
BAB II.pdf Restricted to Repository UPNVJ Only Download (887kB) |
|
Text
BAB III.pdf Restricted to Repository UPNVJ Only Download (1MB) |
|
Text
BAB IV.pdf Restricted to Repository UPNVJ Only Download (2MB) |
|
Text
BAB V.pdf Restricted to Repository UPNVJ Only Download (88kB) |
|
Text
DAFTAR PUSTAKA.pdf Download (130kB) |
|
Text
DAFTAR RIWAYAT HIDUP.pdf Restricted to Repository UPNVJ Only Download (107kB) |
|
Text
LAMPIRAN.pdf Restricted to Repository UPNVJ Only Download (1MB) |
|
Text
HASIL PLAGIARISME.pdf Restricted to Repository staff only Download (41MB) |
|
Text
ARTIKEL KI.pdf Restricted to Repository staff only Download (290kB) |
Abstract
In the context of the case study, Work-in-Progress (WIP) refers to production output that has not reached the Prime category but can still be reprocessed. This study aims to design and evaluate a WIP prediction and monitoring prototype for steel manufacturing using a decoupled machine learning architecture. The research adopts the Design Research Methodology and processes semi-structured production reports through manual-assisted preprocessing, date–profile observation construction, daily WIP allocation, data leakage prevention, and chronological data splitting. Five regression models were compared, with Extra Trees selected as the best-performing model, achieving an RMSE of 177.37 tons, MAE of 110.00 tons, and R² of 0.102 on 24 test observations, while outperforming the mean and median baselines. The model was operationalized through an adapter layer, API services, an asynchronous task queue, an inference worker, model artifacts, a database, and a web interface. Functional testing covered data contracts and adapters, the asynchronous prediction lifecycle, model traceability, and reconciliation. Limited validation by three practitioners produced an average score of 3.86 out of 4. The prototype successfully operationalized the WIP prediction model, while predictions remain preliminary estimates rather than the sole basis for operational decisions.
| Item Type: | Thesis (Skripsi) |
|---|---|
| Additional Information: | [No.Panggil: 2210511145] [Pembimbing: Dr. Noor Falih, S.Kom., M.T] [Penguji 1: Jayanta, S.Kom, M.Si.] [Penguji 2: Muhammad Panji Muslim, S.Pd., M.Kom.] |
| Uncontrolled Keywords: | Work-in-Progress; steel manufacturing; machine learning; decoupled architecture; Extra Trees. |
| Subjects: | H Social Sciences > HD Industries. Land use. Labor Q Science > QA Mathematics > QA75 Electronic computers. Computer science Q Science > QA Mathematics > QA76 Computer software |
| Divisions: | Fakultas Ilmu Komputer > Program Studi Informatika (S1) |
| Depositing User: | MUHAMMAD ARDY CAHYA |
| Date Deposited: | 30 Jul 2026 02:16 |
| Last Modified: | 30 Jul 2026 02:16 |
| URI: | http://repository.upnvj.ac.id/id/eprint/51910 |
Actions (login required)
 |
View Item |