Maulidina, Salsabila Rizka (2025) Penerapan Metode XGBOOST untuk Prediksi Kebutuhan Material pada Studi Kasus Pembuatan KRL KCI di PT INKA. Other thesis, Institut Teknologi Sepuluh Nopember.
![[thumbnail of 5002211150-Undergraduate_Thesis.pdf]](http://repository.its.ac.id/style/images/fileicons/text.png) |
Text
5002211150-Undergraduate_Thesis.pdf - Accepted Version Restricted to Repository staff only Download (4MB) | Request a copy |
Abstract
Pengelolaan kebutuhan material merupakan salah satu aspek yang penting dalam proses manufaktur, termasuk pada produksi Kereta Rel Listrik (KRL) KCI oleh PT Industri Kereta Api (INKA). Permasalahan yang dapat muncul dalam pengelolaan material adalah ketidakseimbangan stok dan ketidakpastian pengadaan. Penelitian ini bertujuan untuk menerapkan metode Extreme Gradient Boosting (XGBoost)sebagai salah satu pembelajaran mesin pada data kebutuhan material KRL KCI di PT INKA. Model dibangun dengan menggunakan data historis kebutuhan material KRL KCI PT. INKA dimana data dibagi menjadi data latih dan data uji dengan rasio 80:20. Pemilihan parameter optimal dilakukan melalui Grid Search terhadap nilai regulasi L1 dan L2, serta dilakukan pelatihan menggunakan skema k-fold cross validation (k = 2 hingga 10) dengan Median Absolute Error (MedAE). Model akhir dibangun menggunakan parameter terbaik dan dievaluasi pada data uji menggunakan beberapa metrik. Hasil menunjukkan performa model yang relatif stabil dengan nilai R2 berkisar antara 0,93 hingga 0,96. Meskipun demikian, ditemukan kecenderungan underestimation pada permintaan besar yang kemungkinan disebabkan oleh ketidakseimbangan distribusi target serta efektransformasilogaritmik.
================================================================================================================================
Material requirements management is an important aspect of the manufacturing process, including in the production of KCI Electric Trains (KRL) by PT Industri Kereta Api (INKA). Problems that can arise in material management are stock imbalances and procurement uncertainty. This study aims to apply the Extreme Gradient Boosting (XGBoost) method as one of the machine learning techniques on KRL KCI material requirement data at PT INKA. The model was built using historical KRL KCI material requirement data from PT INKA, where the data was divided into training and testing data with an 80:20 ratio. The selection of optimal parameters was performed using Grid Search on the values of 80:20. Optimal parameter selection was performed through Grid Search on L1 and L2 regularization values, and training was conducted using a k-fold cross validation scheme (k = 2 to 10), with Median Absolute Error (MedAE) as the primary evaluation metric. The final model was built using the best parameters and evaluated on the test data using several metrics. The results show relatively stable model performance with R2 values ranging from 0.93 to 0.96. However, a tendency toward underestimation was found for large demand, which is likely caused by the imbalance in the target distribution and the effect of logarithmic transformation.
| Item Type: | Thesis (Other) |
|---|---|
| Uncontrolled Keywords: | Kebutuhan Material, KRL KCI, XGBoost |
| Subjects: | Q Science > Q Science (General) > Q325.5 Machine learning. Support vector machines. Q Science > QA Mathematics > QA336 Artificial Intelligence T Technology > TS Manufactures > TS176 Manufacturing engineering. Process engineering (Including manufacturing planning, production planning) |
| Divisions: | Faculty of Mathematics and Science > Mathematics > 44201-(S1) Undergraduate Thesis |
| Depositing User: | Salsabila Rizka Maulidina |
| Date Deposited: | 09 Sep 2025 04:16 |
| Last Modified: | 09 Sep 2025 04:16 |
| URI: | http://repository.its.ac.id/id/eprint/126789 |
Available Versions of this Item
- Penerapan Metode XGBOOST untuk Prediksi Kebutuhan Material pada Studi Kasus Pembuatan KRL KCI di PT INKA. (deposited 09 Sep 2025 04:16) [Currently Displayed]
Actions (login required)
 |
View Item |