Prediksi Harga Dan Stok Komoditas Pangan Menggunakan Extreme Gradient Boosting(XGB) Dengan Whale Optimization Algorithm (WOA)(Studi Kasus Komoditas Jagung)

Ni'mah, Faridatun (2025) Prediksi Harga Dan Stok Komoditas Pangan Menggunakan Extreme Gradient Boosting(XGB) Dengan Whale Optimization Algorithm (WOA)(Studi Kasus Komoditas Jagung). Other thesis, Institut Teknologi Sepuluh Nopember.

Text 5002211025-Undergraduate_Thesis.pdf - Accepted Version Restricted to Repository staff only Download (5MB) | Request a copy

Abstract

Prediksi stok jagung secara akurat merupakan aspek krusial dalam perencanaan ketahanan pangan dan pengelolaan rantai pasok. Namun, fluktuasi stok yang tajam akibat interpolasi data bulanan ke harian sering kali menghasilkan lonjakan tidak realistis yang dapat mengganggu kinerja model prediksi. Penelitian ini mengusulkan pendekatan kombinasi Extreme Gradient Boosting (XGBoost) dengan Whale Optimization Algorithm (WOA) untuk meningkatkan akurasi prediksi stok jagung harian di Indonesia. Sebagai langkah awal, dilakukan koreksi terhadap data stok untuk menghasilkan pola yang lebih realistis melalui simulasi dinamika produksi dan kebutuhan harian, sehingga mengurangi efek lonjakan artifisial. Model XGBoost digunakan sebagai algoritma prediksi utama, sementara WOA dioptimalkan untuk mencari kombinasi hyperparameter terbaik secara adaptif, guna meminimalkan kesalahan prediksi. Evaluasi dilakukan dengan memvariasikan jumlah lag historis sebagai input model, dan kinerja dinilai menggunakan metrik RMSE, MAE, dan MAPE. Hasil eksperimen menunjukkan bahwa model XGB-WOA mampu menghasilkan prediksi yang akurat dan stabil, dengan nilai RMSE terbaik sebesar 5.527 dan MAPE terbaik sebesar 1.55% pada konfigurasi lag 1. Analisis feature importance mengungkapkan bahwa fitur historis seperti Stok_lag_1 dan Stok_MA_7 merupakan kontributor dominan terhadap performa model. Dengan demikian, pendekatan XGB-WOA terbukti efektif dalam memprediksi stok jagung secara akurat, serta memiliki potensi besar untuk diimplementasikan dalam sistem informasi manajemen stok berbasis data time series.
========================================================================================================================
Accurate prediction of daily corn stock is crucial for food security planning and supply chain management. However, sharp fluctuations in stock data, resulting from linear interpolation of monthly data into daily values, often produce unrealistic spikes that can degrade model performance. This study proposes a hybrid approach combining Extreme Gradient Boosting (XGBoost) with the Whale Optimization Algorithm (WOA) to improve the accuracy of daily corn stock forecasting in Indonesia. As a preliminary step, the stock data was corrected to generate a more realistic pattern by simulating daily production and demand dynamics, thereby reducing artificial jumps. XGBoost was employed as the primary predictive algorithm, while WOA was used to optimize hyperparameters adaptively, minimizing prediction error. The evaluation was conducted by varying the number of historical lag features as model inputs, and performance was assessed using RMSE, MAE, and MAPE metrics. Experimental results demonstrate that the XGB-WOA model achieves accurate and stable predictions, with the best RMSE of 5.527 and the best MAPE of 1.55% at lag 1. Feature importance analysis reveals that historical features such as Stok_lag_1 and Stok_MA_7 are the dominant contributors to model performance. Thus, the XGB-WOA approach proves effective in accurately predicting corn stock and has significant potential for implementation in time-series-based stock management information systems.

Item Type:	Thesis (Other)
Uncontrolled Keywords:	prediksi, XGBoost, WOA, feature importance, RMSE, prediction, XGBoost, WOA, feature importance, RMSE
Subjects:	Q Science > QA Mathematics > QA276 Mathematical statistics. Time-series analysis. Failure time data analysis. Survival analysis (Biometry) Q Science > QA Mathematics > QA278.2 Regression Analysis. Logistic regression Q Science > QA Mathematics > QA76.6 Computer programming.
Divisions:	Faculty of Mathematics, Computation, and Data Science > Mathematics > 44201-(S1) Undergraduate Thesis
Depositing User:	Faridatun Ni'mah
Date Deposited:	04 Aug 2025 08:33
Last Modified:	04 Aug 2025 08:33
URI:	http://repository.its.ac.id/id/eprint/127195

Actions (login required)

View Item