Pengembangan Model Integrasi Peramalan Permintaan Dan Kebijakan Persediaan Untuk Produk Sprei

Nurhuda, Dwi Septa (2026) Pengembangan Model Integrasi Peramalan Permintaan Dan Kebijakan Persediaan Untuk Produk Sprei. Masters thesis, Institut Teknologi Sepuluh Nopember.

Text 6032241075-Master_Thesis.pdf - Accepted Version Restricted to Repository staff only Download (2MB) | Request a copy

Abstract

Industri tekstil khusunya produk sprei, menghadapi ketidakpastian permintaan yang tinggi diakibatkan beberapa faktor seperti variasi ukuran, desain atau motif, dan faktor musiman dan beresiko terjadinya overstock atau stockout. PT XYZ saat ini perencaan produksi masih dilakukan berbasis intuisi tanpa metode peramalan baku. Penelitian ini bertujuan mengembangkan model integrasi peramalan permintaan dan kebijakan persediaan untuk tujuh varian ukuran sprei di PT XYZ melalui klasifikasi pola permintaan, pemilihan metode peramalan time series dan penentuan parameter persediaan berbasis continuous review system . Data penjualan mingguan periode Agustus 2024 – Juni 2025 (52 minggu) dieksplorasi menggunakan Exploratory Data Analysis (EDA) meliputi visualisasi time series, uji stasioneritas Augmented Dickey-Fuller (ADF), dan analisa ACF/PACF. Klasifikasi pola permintaan menggunakan metode Syntetos-Boylan berdasarkan Average Demand Interval (ADI) dan Coefficient of Variation Squared (CV2) dan hasil klasifikasi didapatkan ketujuh varian produk sprei masuk dalam kategori smooth Demand (ADI = 1 dan CV2 = 0,10 – 0,25), sehingga metode peramalan ARIMA dan SES diterapkan dan dibandingkan menggunakan skema Rolling Origin Cross-Validation, dievaluasi dengan MAPE, MAE dan RMSE selanjutnya dilakukan uji signifikansi dengan Diebold-Mariano Test. Hasil penelitian menunjukkan SES terpilih sebagai model terbaik untuk empat varian (MAPE 17,16% - 30,03%) dan ARIMA untuk tiga varian lainnya dengan T2 140 B2 menunjukkan perbedaan akurasi yang signifikan secara statistik (p-value = 0,037). Parameter persediaan yang dihasilkan bervariasi untuk safety stock 106pc sampai 284pc dan reorder point 335pc sampai 839pc dengan target service level 95%. Simulasi kebijakan usulan pada varian T2 140 B2 terjadi penurunan rata – rata stok akhir mingguan sebesar 65% dibandingkan kondisi eksisting dengan tetap menjaga service level 95%.
========================================================================================================================================
The textile industry, particularly regarding bed linen products, faces high Demand uncertainty due to variations in size, motifs, promotions, and seasonal factors, Thereby creating risks of overstocking and stockouts. Currently, production planning at PT XYZ is still based on intuition without any standardized Forecasting method. This study aims to develop an integrated model of Demand Forecasting and inventory policy for seven variations of bedsheet sizes at PT XYZ through Demand pattern classification, selection of time series Forecasting methods, and determination of inventory parameters based on a continuous review system . Weekly sales data for the period of August 2024 – June 2025 (52 weeks) were explored using Exploratory Data Analysis (EDA) including time series visualization, Augmented Dickey-Fuller (ADF) stationarity test, and ACF/PACF analysis. Demand pattern classification was conducted using The Syntetos-Boylan method based on The Average Demand Interval (ADI) and Coefficient of Variation Squared (CV^2). The classification results indicated that all seven bedsheet product variants fell into The smooth Demand category (ADI = 1 and CV^2 = 0.10 – 0.25). Consequently, ARIMA and SES Forecasting methods were applied and compared using The Rolling Origin Cross-Validation scheme, evaluated with MAPE, MAE, and RMSE, followed by a significance test using The Diebold-Mariano Test. The results showed that SES was selected as the best model for four variants (MAPE 17.16% - 30.03%) and ARIMA for The other three variants, with T2 140 B2 showing a statistically significant difference in accuracy (p-value = 0.037). The resulting inventory parameters varied, with safety stock ranging from 106 pcs to 284 pcs and reorder points ranging from 335 pcs to 839 pcs at a target service level of 95%. Simulation of the proposed policy on The T2 140 B2 variant showed a 65% reduction in the average weekly ending stock compared to The existing condition, while maintaining a 95% service level.

Item Type:	Thesis (Masters)
Uncontrolled Keywords:	Demand Forecasting, Inventory Management, Time series Analysis, Safety stock, Reorder point, Industri Tekstil, Textile Industry
Subjects:	H Social Sciences > HD Industries. Land use. Labor > HD55 Inventory control
Divisions:	Interdisciplinary School of Management and Technology (SIMT) > 61101-Master of Technology Management (MMT)
Depositing User:	Dwi Septa Nurhuda
Date Deposited:	30 Jul 2026 15:47
Last Modified:	30 Jul 2026 15:47
URI:	http://repository.its.ac.id/id/eprint/140113

Actions (login required)

View Item