Kustia, Puspa Ayu (2026) Komparasi Penerapan Metode Exponential Smoothing, Arima, Dan Random Forest Regression Untuk Peramalan Kebutuhan Dan Perencanaan Persediaan Material Distribusi Utama (MDU) Di PT PLN (Persero) UID Jaya. Masters thesis, Institut Teknologi Sepuluh Nopember.
![[thumbnail of 6010241078-Master_Thesis.pdf]](http://repository.its.ac.id/style/images/fileicons/text.png) |
Text
6010241078-Master_Thesis.pdf - Accepted Version Restricted to Repository staff only Download (8MB) | Request a copy |
Abstract
Peramalan kebutuhan dan perencanaan persediaan Material Distribusi Utama (MDU) merupakan bagian penting dalam pengelolaan rantai pasok ketenagalistrikan untuk menjaga keandalan sistem distribusi, menjamin kontinuitas pelayanan pelanggan, serta meningkatkan efektivitas pengelolaan material di PT PLN (Persero) UID Jaya. Perencanaan yang masih bergantung pada pendekatan historis sederhana dan pertimbangan manual berpotensi menghasilkan ketidaksesuaian antara kebutuhan aktual dan ketersediaan persediaan, sehingga meningkatkan risiko stockout maupun overstock. Penelitian ini membandingkan kinerja tiga metode, yaitu Exponential Smoothing, ARIMA, dan Random Forest Regression, dalam memodelkan peramalan kebutuhan MDU pada material kWh Meter dan MCB, menggunakan data historis pemakaian mingguan tahun 2021–2025 sebanyak 261 observasi, dengan pembagian data Training (80%) dan Testing (20%). Identifikasi awal menunjukkan bahwa data kebutuhan kedua material tidak memiliki tren maupun pola musiman yang signifikan secara praktis, sehingga model non-musiman dinilai lebih sesuai digunakan. Kinerja model dievaluasi menggunakan MAE, MAPE, RMSE, dan WAPE. Random Forest Regression menghasilkan kinerja terbaik untuk kWh Meter (MAE: 6.127,81; MAPE: 78,44%; RMSE: 8.123,08; WAPE: 39,81%), sedangkan Simple Exponential Smoothing menghasilkan kinerja terbaik untuk MCB (MAE: 2.613,49; MAPE: 162,21%; RMSE: 4.225,45; WAPE: 60,47%). Model terbaik pada masing-masing material digunakan sebagai dasar perhitungan Safety Stock dan Reorder Point dalam penyusunan kebijakan persediaan. Simulasi kebijakan persediaan berbasis hasil peramalan menghasilkan peningkatan efisiensi persediaan, ditandai dengan peningkatan Inventory Turnover (ITO) dan penurunan Days of Inventory (DOI) pada kedua material. Penelitian ini memberikan kontribusi ilmiah dalam pemilihan metode peramalan yang sesuai dengan karakteristik data MDU, serta kontribusi praktis berupa rekomendasi perencanaan persediaan yang lebih objektif, terukur, dan berbasis data bagi PT PLN (Persero) UID Jaya.
======================================================================================================================================
Demand forecasting and inventory planning for Main Distribution Materials (MDU) constitute an essential part of electricity distribution supply chain management in maintaining the reliability of the distribution system, ensuring continuity of customer service, and improving the effectiveness of material management at PT PLN (Persero) UID Jaya. Planning that still relies on simple historical approaches and manual judgment has the potential to create discrepancies between actual demand and inventory availability, thereby increasing the risk of both stockout and overstock. This study compares the performance of three methods, namely Exponential Smoothing, ARIMA, and Random Forest Regression, in modeling MDU demand forecasting for kWh Meter and MCB materials, using historical weekly usage data from 2021–2025 comprising 261 observations, divided into Training (80%) and Testing (20%) sets. Preliminary identification indicates that the demand data for both materials exhibit no practically significant trend or seasonal pattern, suggesting that non-seasonal models are more appropriate. Model performance was evaluated using MAE, MAPE, RMSE, and WAPE. Random Forest Regression yielded the best performance for kWh Meter (MAE: 6,127.81; MAPE: 78.44%; RMSE: 8,123.08; WAPE: 39.81%), while Simple Exponential Smoothing yielded the best performance for MCB (MAE: 2,613.49; MAPE: 162.21%; RMSE: 4,225.45; WAPE: 60.47%). The best-performing model for each material was subsequently used as the basis for calculating Safety Stock and Reorder Point in developing an inventory policy. Simulation of the forecast-based inventory policy resulted in improved inventory efficiency, indicated by an increase in Inventory Turnover (ITO) and a decrease in Days of Inventory (DOI) for both materials. This study contributes scientifically to the selection of forecasting methods appropriate to data characteristics, as well as practically by providing more objective, measurable, and data-driven inventory planning recommendations for PT PLN (Persero) UID Jaya.
| Item Type: | Thesis (Masters) |
|---|---|
| Uncontrolled Keywords: | Peramalan, MDU, Exponential Smoothing, ARIMA, Random Forest, Forecasting |
| Subjects: | T Technology > TS Manufactures > TS161 Materials management. |
| Divisions: | Faculty of Industrial Technology and Systems Engineering (INDSYS) > Industrial Engineering > 26101-(S2) Master Thesis |
| Depositing User: | Puspa Ayu Kustia |
| Date Deposited: | 06 Aug 2026 05:57 |
| Last Modified: | 06 Aug 2026 05:57 |
| URI: | http://repository.its.ac.id/id/eprint/144137 |
Actions (login required)
 |
View Item |