Please use this identifier to cite or link to this item:
http://repository.ipb.ac.id/handle/123456789/168604| Title: | Model Prediksi Permintaan Lapis Bogor ABC Menggunakan Algoritma XGBoost dan Random Forest di PT XYZ |
| Other Titles: | Demand Forecasting Model for Lapis Bogor ABC Using XGBoost and Random Forest Algorithms at XYZ Co. |
| Authors: | Hardjomidjojo, Hartrisari Marimin HIKMAH, NUR |
| Issue Date: | 2025 |
| Publisher: | IPB University |
| Abstract: | Peramalan permintaan produk Lapis Bogor ABC dengan umur simpan pendek di PT XYZ menghadapi tantangan akurasi, menyebabkan masalah operasional seperti overstock dan stockout akibat ketidakmampuan metode konvensional dalam memodelkan pola permintaan yang fluktuatif. Penelitian ini bertujuan mengembangkan dan membandingkan model prediksi berbasis machine learning untuk meningkatkan akurasi peramalan pada dua varian produk Lapis Bogor ABC. Menggunakan algoritma Random Forest dan XGBoost, penelitian ini menganalisis data historis melalui rekayasa fitur dan optimisasi hyperparameter. Hasil analisis menunjukkan bahwa faktor musiman terkait hari raya besar, khususnya Lebaran, merupakan prediktor permintaan paling dominan. Evaluasi komparatif menemukan tidak ada satu model pun yang unggul secara universal, model XGBoost mencapai akurasi tertinggi untuk varian original keju yaitu sebesar 88,13%, sementara Random Forest lebih superior untuk varian brownies keju sebesar 90,16%. Hal ini secara konklusif memvalidasi bahwa pendekatan peramalan one-size-fits-all tidak efektif. Pemilihan model prediktif yang disesuaikan secara spesifik dengan karakteristik unik setiap produk menjadi krusial untuk optimisasi dan pengambilan keputusan yang lebih akurat. Demand forecasting for short shelf life Lapis Bogor ABC products at PT XYZ faces accuracy challenges, leading to operational issues such as overstock and stockouts due to the inability of conventional methods to model fluctuating demand patterns. This study aims to develop and compare machine learning based predictive models to improve forecasting accuracy for two variants of the Lapis Bogor ABC product. Using the Random Forest and XGBoost algorithms, the study analyzes historical data through feature engineering and hyperparameter optimization. The analysis reveals that seasonal factors related to major holidays, particularly Lebaran, are the most dominant demand predictors. Comparative evaluation shows that no single model universally outperforms the other. XGBoost achieved the highest accuracy for the original cheese variant at 88.13%, while Random Forest performed better for the cheese brownies variant at 90.16%. These findings conclusively validate that a one size fits all forecasting approach is ineffective. Selecting predictive models specifically tailored to the unique characteristics of each product is crucial for optimization and more accurate decision making. |
| URI: | http://repository.ipb.ac.id/handle/123456789/168604 |
| Appears in Collections: | UT - Agroindustrial Technology |
Files in This Item:
| File | Description | Size | Format | |
|---|---|---|---|---|
| Cover-Nur Hikmah.pdf | Cover | 4.02 MB | Adobe PDF | View/Open |
| FullText-Nur Hikmah.pdf Restricted Access | Fulltext | 3.12 MB | Adobe PDF | View/Open |
| Lampiran-Nur Hikmah.pdf Restricted Access | Lampiran | 1.83 MB | Adobe PDF | View/Open |
Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.