Analisis Komparatif Model Regresi untuk Mengidentifikasi Faktor Pengaruh Produksi Minyak Sawit Mentah

Nurtaniyahya, Ilham (2026) Analisis Komparatif Model Regresi untuk Mengidentifikasi Faktor Pengaruh Produksi Minyak Sawit Mentah. Masters thesis, Institut Teknologi Sepuluh Nopember.

Text 6032241090-Master_Thesis.pdf - Accepted Version Restricted to Repository staff only Download (2MB) | Request a copy

Abstract

Industri kelapa sawit merupakan sektor penting dalam perekonomian Indonesia, khususnya dalam produksi Crude Palm Oil (CPO). Namun, industri ini menghadapi tantangan fluktuasi produksi CPO yang berdampak pada efisiensi operasional dan profitabilitas. Penelitian ini bertujuan mengidentifikasi faktor yang berpengaruh terhadap produksi CPO di salah satu industri pengolahan kelapa sawit di Kalimantan Barat. Metodologi penelitian mengikuti kerangka kerja CRISP-DM dengan teknik pemodelan regresi yaitu RFR, XGBoost, SVR, dan LR melalui lima skenario seleksi fitur, yaitu penggunaan seluruh fitur, pengecualian multikolinearitas, pemilihan 10 fitur penting, seleksi fitur berdasarkan PFI, dan penggabungan skenario pengecualian multikolinearitas dengan pemilihan 10 fitur penting. Hasil evaluasi model menggunakan metrik MAE, RMSE, MAPE, dan R^2 menunjukkan bahwa model RFR secara konsisten lebih unggul dibandingkan ketiga model lainnya dengan performa terbaik diperoleh pada skenario kelima dengan nilai MAE sebesar 19.164 kg, RMSE sebesar 26.138 kg, MAPE sebesar 14,22%, dan R^2 sebesar 0,58. Berdasarkan hasil evaluasi model dan skenario tersebut, faktor yang berpengaruh terhadap produksi CPO adalah penerimaan TBS, TBS restan kemarin, sortasi, FFA, kapasitas olah, oil losses, curah hujan lag_1, dan curah hujan. Sementara itu, faktor yang tidak berpengaruh adalah mentah dan abnormal. Hasil wawancara dengan lima ahli di lapangan menunjukkan bahwa seluruh variabel penelitian berpengaruh terhadap produksi CPO. Tingkat kepentingan kelima ahli ditentukan melalui pembobotan AHP, dengan asisten proses 1 sebagai ahli dengan pembobotan tertinggi.
=======================================================================================================================================
The palm oil industry is an important sector in the Indonesian economy, especially in the Crude Palm Oil (CPO) production. However, the industry faces the challenges of fluctuations in CPO production, which affect its operational efficiency and profitability. This study aimed to identify the factors that significantly influence CPO production at a palm oil processing industry in West Kalimantan. The research methodology based on CRISP-DM framework with regression modeling techniques, namely RFR, XGBoost, SVR, and LR, across five different feature selection scenarios: the use of all features, exclusion of multicollinearity, selection of the 10 most important features, feature selection based on PFI, and combination of the multicollinearity exclusion scenario with the selection of the 10 most important features. The model evaluation results using the MAE, RMSE, MAPE, and R² metrics showed that the RFR model consistently outperformed the other three models, with the best performance obtained in the fifth scenario, yielding an MAE of 19.164 kg, an RMSE of 26.138 kg, an MAPE of 14,22%, and an R^2 of 0,58. Based on the results of the model and scenario evaluation, the influencing factors CPO production were FFB reception, FFB remaining yesterday, sorting, FFA, processing capacity, oil losses, rainfall lag_1, and rainfall. Meanwhile, the factors found to have no influence were unripe and abnormal. The results of interviews with five experts in the field show that all research variables affect CPO production. The level of importance of the five experts is determined through AHP weighting, with process assistant 1 as the expert with the highest weighting.

Item Type:	Thesis (Masters)
Uncontrolled Keywords:	Pengaruh produksi, CPO, CRISP-DM, Production influence
Subjects:	T Technology > T Technology (General) > T57.5 Data Processing
Divisions:	Interdisciplinary School of Management and Technology (SIMT) > 61101-Master of Technology Management (MMT)
Depositing User:	Ilham Nurtaniyahya
Date Deposited:	29 Jul 2026 08:24
Last Modified:	29 Jul 2026 08:24
URI:	http://repository.its.ac.id/id/eprint/140575

Actions (login required)

View Item