Detail Cantuman

Harga ikan memiliki peran vital dalam mendukung pembangunan sosial ekonomi dan aktivitas industri perikanan. Namun, harga ikan sering berfluktuasi akibat faktor musiman, kondisi cuaca, serta dinamika pasar. Untuk mengatasi hal tersebut, penelitian ini menerapkan model Seasonal Autoregressive Integrated Moving Average (SARIMA) yang mampu menangkap pola musiman pada data deret waktu, serta mengintegrasikan Bayesian Optimization dalam pemilihan parameter. Keunggulan Bayesian Optimization terletak pada kemampuannya melakukan pencarian parameter secara otomatis tanpa perlu trial and error, sekaligus efisien dalam menghemat waktu dan sumber daya komputasi dibandingkan metode konvensional.
Hasil penelitian menunjukkan bahwa metode SARIMA menghasilkan rata-rata nilai Mean Absolute Percentage Error (MAPE) sebesar 29,86%, sedangkan model dengan Bayesian Optimization mampu menurunkan rata-rata MAPE menjadi 19,57%. Selain itu, hasil prediksi harga ikan tahun 2025 menunjukkan variasi antar jenis komoditas. Sebagai contoh, harga prediksi SARIMA yang sudah dioptimasi oleh Bayesian untuk Bandeng berada pada kisaran Rp53.885,72 per kilogram, Cakalang sebesar Rp48.463,04, dan Cumi-cumi mencapai Rp83.864,67. Sementara itu, harga Kakap Merah diprediksi sebesar Rp70.267,97, Kerapu Rp51.574,24, dan Merah Rp130.930,89 per kilogram.
Dengan demikian, integrasi SARIMA dan Bayesian Optimization terbukti dapat meningkatkan akurasi peramalan harga ikan pada sebagian besar komoditas. Pendekatan ini diharapkan dapat memberikan kontribusi dalam pengembangan metode prediksi harga komoditas perikanan yang lebih reliabel, serta menjadi dasar pengambilan keputusan strategis bagi para pemangku kepentingan di sektor perikanan.

Fish prices play a vital role in supporting socio-economic development and the activities of the fisheries industry. However, fish prices often fluctuate due to seasonal factors, weather conditions, and market dynamics. To address this issue, this study applies the Seasonal Autoregressive Integrated Moving Average (SARIMA) model, which is capable of capturing seasonal patterns in time series data, and integrates Bayesian Optimization for parameter selection. The advantage of Bayesian Optimization lies in its ability to search for optimal parameters automatically without requiring trial and error, while efficiently saving time and computational resources compared to conventional methods.
The results indicate that the SARIMA method yields an average Mean Absolute Percentage Error (MAPE) of 29.86%, whereas the model with Bayesian Optimization reduces the average MAPE to 19.57%. Furthermore, the predicted fish prices for 2025 show variations across commodity types. For instance, the Bayesian-optimized SARIMA predictions indicate prices of approximately IDR 53,885.72 per kilogram for milkfish (Bandeng), IDR 48,463.04 for skipjack tuna (Cakalang), and IDR 83,864.67 for squid (Cumi-cumi). Meanwhile, red snapper (Kakap Merah) is predicted at IDR 70,267.97, grouper (Kerapu) at IDR 51,574.24, and tuna (Merah) at IDR 130,930.89 per kilogram.
In conclusion, the integration of SARIMA and Bayesian Optimization has proven to improve the accuracy of fish price forecasting for most commodities. This approach is expected to contribute to the development of more reliable fish commodity price prediction methods and serve as a basis for strategic decision-making by stakeholders in the fisheries sector.