Demand Forecasting Menggunakan Metode Arima dan Neural Network pada Industri Pengolahan Susu : Studi Kasus PT ABC Kogen Dairy

Harsono, Ardhiansyah Widhi (2023) Demand Forecasting Menggunakan Metode Arima dan Neural Network pada Industri Pengolahan Susu : Studi Kasus PT ABC Kogen Dairy. Masters thesis, Institut Teknologi Sepuluh Nopember.

Text 6032202061-Master_Thesis.pdf - Accepted Version Restricted to Repository staff only until 1 October 2025. Download (1MB) | Request a copy

Abstract

Operasional rantai pasok perusahaan banyak dihadapkan pada situasi ketidakpastian yang kompleks, seperti ketidakpastian permintaan, ketidakpastian ekonomi, pandemi, dan ketidakpastian pasokan bahan baku. Dengan meningkatnya kompleksitas masalah yang dihadapi, diperlukan aliran informasi demand forecast yang baik untuk dapat mencapai keberhasilan rantai pasok. Karakteristik industri pengolahan susu adalah fleksibilitas bahan baku utama rendah, dan umur produk yang pendek. Dengan karakteristik tersebut, diperlukan demand forecast yang akurat agar operational berjalan efektif dan efisien. Metode forecasting saat ini memiliki tingkat akurasi demand forecast di bawah 80% pada tahun 2022.
Pada penelitian ini akan dilakukan forecasting dengan metode ARIMA dan neural network dengan periode forecasting 1 bulan, 2 bulan, dan 3 bulan pada masing-masing SKU. Mean Absolute Percentage Error (MAPE) digunakan untuk pengukuran akurasi demand forecast yang dihasilkan.
Dari hasil penelitian. pada SKU KBY Original, ARIMA memiliki nilai MAPE terkecil 19.87%, dan akurat pada periode forecasting 1 bulan, ANN memiliki nilai MAPE terkecil 9.96% dan akurat hingga periode forecasting 3 bulan. Pada SKU KBY Strawberry, ANN memiliki nilai MAPE terkecil 13.79% dan akurat hanya pada periode forecasting 1 bulan. Pada SKU KBY Blueberry, ARIMA memiliki nilai MAPE terkecil 19.08%, dan akurat hingga periode forecasting 3 bulan, ANN memiliki nilai MAPE terkecil 16.21% dan akurat hingga periode forecasting 3 bulan. Pada SKU Paseteurized Fullcream, ANN memiliki nilai MAPE terkecil 11.29% dan akurat hingga periode forecasting 3 bulan. Pada SKU Pasteurized Chocolate, ARIMA memiliki nilai MAPE terkecil 18.49%, dan akurat hingga periode forecasting 3 bulan, ANN memiliki nilai MAPE terkecil 16.39% dan hanya akurat pada periode forecasting 1 bulan. Secara umum metode terbaik pada penelitian ini adalah ANN, dimana ANN akurat pada 5 SKU sedangkan ARIMA akurat hanya pada 3 SKU. Periode forecasting terbaik adalah 1 bulan yang memberikan nilai MAPE terkecil.
====================================================================================================================================
Many companies' supply chain operations are faced with complex uncertainty situations, such as demand uncertainty, economic uncertainty, pandemics, and raw material supply uncertainty. With the increasing complexity of the problems faced, a good flow of information od femand forecast is needed to be able to achieve supply chain success. The characteristics dairy industry is the main raw material flexibility is low, short product life. The current forecasting method has a demand forecast accuracy rate below 80% in 2022.
In this study, forecasting will be carried out with ARIMA and neural network models with time horizon of 1 month, 2 months, and 3 months. Mean Absolute Percentage Error (MAPE) is used to measure the accuracy to identify what kind of forecasting metod and time horizon that suitable to each SKU.
From the research results, on the KBY Original SKU, the accurate method is ARIMA which has the smallest MAPE value of 19.87%, and is accurate on 1 month forecasting period, the ANN method has the smallest MAPE value of 9.96% and accurate up to a 3 month forecasting period. In the KBY Strawberry SKU, the most accurate method is ANN which has the smallest MAPE value of 13.79% and accurate only in the 1-month forecasting period. In the Blueberry KBY SKU, the accurate method is ARIMA which has the smallest MAPE value of 19.08%, and accurate up to a 3-month forecasting period, the ANN method has the smallest MAPE value of 16.21% and is accurate up to a 3-month forecasting period. In Pasteurized Fullcream SKUs, the most accurate method is the ANN method which has the lowest MAPE value of 11.29% and is accurate up to a 3-month forecasting period. In Pasteurized Chocolate SKUs, the accurate method is ARIMA which has the smallest MAPE value of 18.49%, and is accurate up to a 3-month forecasting period, the ANN method has the smallest MAPE value of 16.39% and only accurate for a 1-month forecasting period. In general, in this research best forecasting method is ANN, which are accurate on 5 SKUs, while ARIMA is only accurate on 3 SKUs. Best time horizon is 1 month, because it gives lowest MAPE.

Item Type:	Thesis (Masters)
Uncontrolled Keywords:	demand forecast, ARIMA, neural network, MAPE
Subjects:	T Technology > T Technology (General) T Technology > TS Manufactures T Technology > TS Manufactures > TS161 Materials management.
Divisions:	Interdisciplinary School of Management and Technology (SIMT) > 61101-Master of Technology Management (MMT)
Depositing User:	Ardhiansyah Widhi Harsono
Date Deposited:	09 Aug 2023 05:34
Last Modified:	09 Aug 2023 05:34
URI:	http://repository.its.ac.id/id/eprint/104451

Actions (login required)

View Item