Intermittent Demand Forecasting and Inventory Service Level Evaluation Using The TSB Method and Machine Learning: A Case Study on PT X in The Document Solutions Industry
Jurnal Locus Penelitian dan Pengabdian · 04.11.2026 · 5 (11) · 9652-9671
DOI: 10.58344/locus.v5i11.6328 ↗Kurz gefasst
Fachartikel in der Zeitschrift Jurnal Locus Penelitian dan Pengabdian, online erschienen am 04.11.2026. Verfasst von Fitrotun Hasanah und 2 weiteren Autor:innen. Der Volltext ist frei zugänglich (Open Access, CC BY-SA 4.0).
Abstract (Originalsprache)
Intermittent demand presents significant challenges in demand forecasting and inventory management, particularly in the document solutions industry, where demand is driven by both rental and sales activities. Forecasting accuracy alone does not necessarily translate into improved inventory performance. This study evaluates the forecasting performance of the Teunter–Syntetos–Babai (TSB) method and machine learning models, including Random Forest, XGBoost, and LightGBM, across total, rental, and sales demand categories and examines their implications for inventory service levels. A quantitative analysis was conducted using supply demand data for 134 stock-keeping units (SKUs) of Brand Z at PT X from August 2023 to November 2025. Demand patterns were classified using the Average Demand Interval (ADI) and the squared coefficient of variation (CV²). Forecasting models incorporated historical demand, lead times, new printer installations, and new printer sales. Forecasting performance was evaluated using root mean squared error (RMSE) and mean absolute error (MAE), while inventory performance was assessed using an order-up-to-level (S) policy based on the probability of no stockout (S1), fill rate (S2), and ready rate (S3). Random Forest demonstrated superior overall performance in terms of forecasting accuracy and inventory service levels, particularly for intermittent and lumpy demand patterns. However, greater forecasting accuracy did not consistently translate into better inventory performance. The TSB method performed better for smooth demand patterns and on several service-level indicators for erratic demand. Demand decomposition also revealed distinct characteristics between rental and sales demand. These findings indicate that forecasting accuracy alone is insufficient for selecting an appropriate forecasting method. Demand characteristics and inventory service-level performance must also be considered to support effective inventory management.
Zitieren
APA
Hasanah, F., Lestiani, M. E., & Maniah, M. (2026). Intermittent Demand Forecasting and Inventory Service Level Evaluation Using The TSB Method and Machine Learning: A Case Study on PT X in The Document Solutions Industry. Jurnal Locus Penelitian dan Pengabdian, 5(11), 9652-9671. https://doi.org/10.58344/locus.v5i11.6328
Vancouver
Hasanah F, Lestiani ME, Maniah M. Intermittent Demand Forecasting and Inventory Service Level Evaluation Using The TSB Method and Machine Learning: A Case Study on PT X in The Document Solutions Industry. Jurnal Locus Penelitian dan Pengabdian. 2026;5(11):9652-9671. doi:10.58344/locus.v5i11.6328
BibTeX
@article{Hasanah2026,
author = {Hasanah, Fitrotun and Lestiani, Melia Eka and Maniah, Maniah},
title = {Intermittent Demand Forecasting and Inventory Service Level Evaluation Using The TSB Method and Machine Learning: A Case Study on PT X in The Document Solutions Industry},
journal = {Jurnal Locus Penelitian dan Pengabdian},
year = {2026},
volume = {5},
number = {11},
pages = {9652--9671},
doi = {10.58344/locus.v5i11.6328}
}
EUVIMED