Forecasting the Price of Staple Food Eggs with XGBoost: A Case Study in the Netherlands

Maulana Putra, Mochamad Rafie Bimantoro (2026) Forecasting the Price of Staple Food Eggs with XGBoost: A Case Study in the Netherlands. Other thesis, Institut Teknologi Sepuluh Nopember.

Text 05211942000017-Undergraduate_Thesis.pdf - Accepted Version Restricted to Repository staff only Download (2MB) | Request a copy

Abstract

Penelitian ini mengembangkan model peramalan harga telur sebagai komoditas pangan pokok menggunakan algoritma Extreme Gradient Boosting (XGBoost) dengan optimasi hyperparameter melalui GridSearchCV. Data yang digunakan berupa harga telur mingguan kategori barn dan cage di Belanda yang diperoleh dari European Commission Agri-Food Data Portal untuk periode 2018–2025. Tahapan penelitian meliputi pra-pemrosesan data, pembentukan fitur deret waktu dengan variasi look-back window, serta analisis sensitivitas terhadap panjang data pelatihan dan horizon peramalan.Hasil penelitian menunjukkan bahwa data harga telur memiliki pola musiman tahunan yang berulang, bukan pola siklus jangka panjang. Konfigurasi terbaik diperoleh dengan look-back window 4 minggu, panjang data pelatihan 2 tahun, dan horizon peramalan 3 bulan. Model cenderung menghasilkan under-forecasting, yang berdampak negatif bagi produsen namun relatif menguntungkan bagi konsumen. Perbandingan model menunjukkan bahwa XGBoost lebih unggul untuk peramalan jangka pendek, sedangkan AdaBoost memiliki potensi dalam menangkap tren jangka panjang. Temuan ini menegaskan bahwa XGBoost merupakan metode yang andal untuk peramalan harga telur jangka pendek hingga menengah di Belanda.
====================================================================================================================================
This study develops an egg price forecasting model as a staple food commodity using the Extreme Gradient Boosting (XGBoost) algorithm with hyperparameter optimization through GridSearchCV. The dataset consists of weekly barn and cage egg prices in the Netherlands obtained from the European Commission’s Agri-Food Data Portal for the 2018–2025 period. The research process includes data preprocessing, time-series feature construction with varying look-back windows, and sensitivity analysis on training data length and forecasting horizon.The results confirm that egg prices exhibit a recurring annual seasonal pattern rather than a long-term cyclical trend. The optimal configuration uses a 4-week look-back window, 2 years of training data, and a 3-month forecasting horizon. The model predominantly produces under forecasting results, which may be unfavorable for producers but beneficial for consumers. Model comparison indicates that XGBoost outperforms AdaBoost in short-term forecasting, while AdaBoost shows potential in capturing longer-term trends. These findings demonstrate that XGBoost is a robust and effective approach for short- to medium-term egg price forecasting in the Netherlands.

Item Type:	Thesis (Other)
Uncontrolled Keywords:	XGBoost, Peramalan Deret Waktu, Harga Telur, Musiman, XGBoost, Time Series Forecasting, Egg Price, Seasonality
Subjects:	T Technology > T Technology (General) > T57.5 Data Processing
Divisions:	Faculty of Intelligent Electrical and Informatics Technology (ELECTICS) > Information System > 57201-(S1) Undergraduate Thesis
Depositing User:	Mochamad Rafie Bimantoro Maulana Putra
Date Deposited:	26 Jan 2026 06:59
Last Modified:	26 Jan 2026 06:59
URI:	http://repository.its.ac.id/id/eprint/130348

Actions (login required)

View Item