Scholarlink

<p>Accurate fish price prediction is crucial to support sustainable fisheries policies and enhance the economic well-being of fishermen amidst challenges such as declining fish stocks due to illegal, unregulated, and unreported (IUU) fishing, and the impacts of climate change. This study aims to predict fish prices at Nizam Zachman Fishing Port using Gated Recurrent Unit (GRU) and Long Shor Term Memory (LSTM) models. The GRU and LSTM methods were employed to analyze input data, including time (date, month, and year), fish species, landing volume, and fish prices for various species such as <i>Loligo Spp</i>,<i> Decapterus macrosoma</i>,<i> Scomberomorus commerson</i>,<i> Selar crumenophthalmus</i>,<i> Thunnus alalunga</i>,<i> Thunnus albacares</i>,<i> Xiphias gladius</i>,<i> Katsuwonus pelamis</i>,<i> Decapterus russelli</i> and <i>Thunnus obesus</i>. The application of GRU and LSTM models is expected to effectively capture the complex and nonlinear characteristics of prices in the fisheries sector. Among the deep learning models, the GRU-5 configuration with k-fold validation demonstrated the best performance, consistently yielding lower MAPE, MAE, and RMSE values compared to the LSTM-5 model. However, the ARMAX model outperformed both LSTM and GRU for several fish species, achieving the lowest MAPE values for instance, 3.23% for <i>Scomberomorus commerson</i> and 3.13% for <i>Katsuwonus pelamis</i>. These findings underscore the robustness of the ARMAX model, particularly for species exhibiting more stable and linear price patterns. Accurate prediction results can be utilized by the Indonesian government to implement measured fishing policies. This approach can help anticipate price uncertainties, promote price stability, and enhance the welfare of fishermen.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Optimization Model for Predicting Fish Prices Using Artificial Intelligence Approach (A Case Study at Nizam Zachman Fishing Port, Jakarta)

  • 期刊论文

  • 发表日期：

    2025年8月12日

查看全文
  • Wirata Wirata,
  • Sugeng Hari Wisudo,
  • Yopi Novita,
  • Mohammad Imron,
  • Yaser Krisnafi

摘要

Accurate fish price prediction is crucial to support sustainable fisheries policies and enhance the economic well-being of fishermen amidst challenges such as declining fish stocks due to illegal, unregulated, and unreported (IUU) fishing, and the impacts of climate change. This study aims to predict fish prices at Nizam Zachman Fishing Port using Gated Recurrent Unit (GRU) and Long Shor Term Memory (LSTM) models. The GRU and LSTM methods were employed to analyze input data, including time (date, month, and year), fish species, landing volume, and fish prices for various species such as Loligo Spp, Decapterus macrosoma, Scomberomorus commerson, Selar crumenophthalmus, Thunnus alalunga, Thunnus albacares, Xiphias gladius, Katsuwonus pelamis, Decapterus russelli and Thunnus obesus. The application of GRU and LSTM models is expected to effectively capture the complex and nonlinear characteristics of prices in the fisheries sector. Among the deep learning models, the GRU-5 configuration with k-fold validation demonstrated the best performance, consistently yielding lower MAPE, MAE, and RMSE values compared to the LSTM-5 model. However, the ARMAX model outperformed both LSTM and GRU for several fish species, achieving the lowest MAPE values for instance, 3.23% for Scomberomorus commerson and 3.13% for Katsuwonus pelamis. These findings underscore the robustness of the ARMAX model, particularly for species exhibiting more stable and linear price patterns. Accurate prediction results can be utilized by the Indonesian government to implement measured fishing policies. This approach can help anticipate price uncertainties, promote price stability, and enhance the welfare of fishermen.