Kebijakan Persediaan Material Distribusi Pada Perusahaan Listrik Dengan Dukungan Peramalan Permintaan Berbasis Machine Learning

Fahrudin, Mahendra Firdausi Danang (2026) Kebijakan Persediaan Material Distribusi Pada Perusahaan Listrik Dengan Dukungan Peramalan Permintaan Berbasis Machine Learning. Other thesis, Institut Teknologi Sepuluh Nopember.

Text 5010221109-Undergraduate_Thesis.pdf - Accepted Version Restricted to Repository staff only Download (11MB) | Request a copy

Abstract

Pengendalian persediaan material distribusi merupakan salah satu tantangan utama bagi perusahaan listrik yang menerapkan sistem distribusi bertingkat. Penelitian ini mengevaluasi penerapan machine learning untuk peramalan permintaan serta menganalisis strategi konsolidasi pengadaan pada tingkat Region dan Wilayah. Penelitian dilakukan pada material MCB, KWH Meter, dan Cable Power di Region Jawa Timur dan Jawa Barat. Tahapan penelitian meliputi klasifikasi pola permintaan menggunakan metode Fast, Slow, and Non-Moving (FSN) dan Average Demand Interval–Coefficient of Variance (ADI–CV), peramalan permintaan menggunakan Long Short-Term Memory (LSTM) dan XGBoost, perancangan kebijakan persediaan Min-Max, serta evaluasi melalui simulasi Monte Carlo pada beberapa skenario konsolidasi. Hasil penelitian menunjukkan bahwa kedua model peramalan mampu mendapat rata-rata MASE di bawah 1, yaitu 0,745 untuk LSTM dan 0,647 untuk XGBoost. XGBoost memberikan performa lebih baik dengan menghasilkan nilai MASE yang lebih rendah pada 9 dari 12 varian material serta waktu komputasi rata-rata 0,12 detik, sekitar 107 kali lebih cepat dibandingkan LSTM. Hasil simulasi menunjukkan bahwa material dengan dominasi holding cost lebih tinggi sesuai dikelola menggunakan strategi Sentralisasi Informasi, sedangkan material dengan dominasi transportation cost lebih efektif menggunakan Sentralisasi Fisik. Hal ini menunjukkan bahwa strategi pengendalian persediaan perlu disesuaikan dengan karakteristik biaya masing-masing material untuk memperoleh efisiensi biaya persediaan tanpa menurunkan tingkat pelayanan.
====================================================================================================================================
Improving inventory control of distribution materials is a key challenge for electric utility companies operating multi-echelon distribution systems. This research evaluates the implementation of machine learning for demand forecasting and analyzes procurement consolidation strategies at the Regional and Area levels. The study focuses on MCB, KWH Meter, and Cable Power materials in the East Java and West Java Regions. The research methodology consists of demand pattern classification using the Fast, Slow, and Non-Moving (FSN) and Average Demand Interval–Coefficient of Variation (ADI–CV) methods, demand forecasting using Long Short-Term Memory (LSTM) and Extreme Gradient Boosting (XGBoost), Min-Max inventory policy design, and policy evaluation through Monte Carlo simulation under several consolidation scenarios. The results indicate that both forecasting models achieved average MASE values below 1, with 0.745 for LSTM and 0.647 for XGBoost, outperforming the naïve baseline. XGBoost demonstrated superior forecasting performance by achieving lower MASE values for 9 out of 12 material variants while requiring an average computation time of only 0.12 seconds, approximately 107 times faster than LSTM. Furthermore, the simulation results reveal that materials with holding cost as the dominant cost component are more effectively managed using the Information Centralization strategy, whereas materials with transportation cost dominance are better managed using the Physical Centralization strategy. These findings demonstrate that inventory control strategies should be tailored to the cost characteristics of each material to improve inventory cost efficiency while maintaining service levels.

Item Type:	Thesis (Other)
Uncontrolled Keywords:	Kebijakan Persediaan, LSTM, Machine Learning, Peramalan, Strategi Konsolidasi, XGBoost, Consolidation Strategy, Forecasting, Inventory Policy, LSTM, Machine Learning, XGBoost.
Subjects:	H Social Sciences > HD Industries. Land use. Labor > HD55 Inventory control T Technology > T Technology (General) > T174 Technological forecasting
Divisions:	Faculty of Industrial Technology and Systems Engineering (INDSYS) > Industrial Engineering > 26201-(S1) Undergraduate Thesis
Depositing User:	Mahendra Firdausi Danang Fahrudin
Date Deposited:	28 Jul 2026 03:16
Last Modified:	28 Jul 2026 03:16
URI:	http://repository.its.ac.id/id/eprint/138349

Actions (login required)

View Item