Prediksi Produksi Perkebunan Besar di Indonesia Menggunakan Metode Category Boosting

Hutama, Faishal Fernando (2027) Prediksi Produksi Perkebunan Besar di Indonesia Menggunakan Metode Category Boosting. Undergraduate thesis, UPN Veteran Jawa Timur.

	Text (Cover) 21081010304-Cover.pdf Download (5MB)
	Text (Bab 1) 21081010304-Bab 1.pdf Download (167kB)
	Text (Bab 2) 21081010304-Bab 2.pdf Restricted to Repository staff only until 4 December 2027. Download (371kB)
	Text (Bab 3) 21081010304-Bab 3.pdf Restricted to Repository staff only until 4 December 2027. Download (658kB)
	Text (Bab 4) 21081010304-Bab 4.pdf Restricted to Repository staff only until 4 December 2027. Download (5MB)
	Text (Bab 5) 21081010304-Bab 5.pdf Download (156kB)
	Text (Daftar pustaka) 21081010304-Daftar pustaka.pdf Download (224kB)
	Text (Lampiran) 21081010304-Lampiran.pdf Restricted to Repository staff only until 4 December 2027. Download (285kB)

Abstract

The plantation subsector plays an important role in Indonesia’s economy by providing non-oil export commodities and absorbing a large workforce. However, its production is strongly influenced by climatic factors such as rainfall and air temperature. This study aims to predict the production of large-scale plantations in Indonesia using the Category Boosting algorithm, an advanced version of the Gradient Boosting Decision Tree (GBDT) that effectively handles categorical data without manual encoding. The data were obtained from BPS and BMKG covering the period 2009–2024, including five main commodities: dry rubber, palm oil, coffee, tea, and sugarcane. The research stages include data preprocessing, handling missing values through linear interpolation, feature engineering (seasonal transformation and lag features), and model training under three scenarios: baseline, hyperparameter optimization using Random Search, and forecasting with Walk-Forward validation. The results show that the Category Boosting model demonstrates strong generalization capability, achieving an R2 value above 0.9 in the best scenario even when tested using a strict Walk-Forward validation method. Furthermore, the model underwent a sensitivity analysis and proved capable of projecting production trends up to 2027 under ±10% climate fluctuation scenarios. These findings indicate that Category Boosting is effective in modeling the complex relationship between climate variables and plantation production, providing a reliable foundation for agricultural policy planning and national food security strategies.

Item Type:

Thesis (Undergraduate)

Contributors:

Contribution	Contributors	NIDN/NIDK	Email
Thesis advisor	Puspaningrum, Eva Yulia	0005078908	evapuspaningrum.if@upnjatim.ac.id
Thesis advisor	Nurlaili, Afina Lina	0013129303	afina.lina.if@upnjatim.com

Subjects:

T Technology > T Technology (General)

Divisions:

Faculty of Computer Science > Departemen of Informatics

Depositing User:

nando Faishal Fernando Hutama

Date Deposited:

04 Dec 2025 06:57

Last Modified:

04 Dec 2025 06:57

URI:

https://repository.upnjatim.ac.id/id/eprint/47819