Market

How AI is revolutionizing fish price prediction in Taiwan

Photo of author

By Milthon Lujan

Graphic summary of the study. Source: Yi-Ting Lai et al., (2024); Aquaculture.
Graphic summary of the study. Source: Yi-Ting Lai et al., (2024); Aquaculture.

The fish and seafood market is a complex ecosystem where price fluctuations can significantly affect everyone involved, from fishermen to consumers. The unpredictable volatility of prices poses a threat to food security, especially for vulnerable populations, and hinders both sustainable fisheries management and economic benefits.

Researchers from Ming Chi University of Technology, National Chengchi University, and Hyson Technology Inc. propose a hybrid fish price prediction model that integrates multiple machine learning and deep learning models to identify and predict fish price dynamics in various wholesale markets.

This article delves into the world of machine learning and deep learning, exploring how they can equip fishermen with the necessary knowledge to make informed decisions for a prosperous future.

The old customs

Traditionally, fishermen relied on experience and observations from different markets to navigate uncertainties. However, accurate predictions are crucial for informed decision-making and maximizing profits while ensuring sustainable practices.

Prices fluctuate greatly due to a confluence of factors: tropical and subtropical weather patterns, fluctuating volumes of fish import/export, and unpredictable political and economic climates. Price fluctuations directly impact:

  • Food security: Fish price volatility can disrupt access to nutritious food, especially for vulnerable populations.
  • Sustainable management: Unpredictable prices make it difficult for fishermen to plan long-term and implement sustainable practices.

The new frontier: machine learning and deep learning

Existing forecasting models, primarily based on econometric and statistical methods, struggle with nonlinear data and multifactorial influences. Additionally, climate change significantly impacts aquaculture systems, causing unforeseen fluctuations and hindering prediction accuracy.

See also  Chefs want to serve up a good story together with farmed cod

Recent advancements in machine learning (ML) and deep learning (DL) offer promising solutions for accurate fish price forecasting. These methods use sophisticated algorithms to analyze data and learn patterns, aiming to achieve superior prediction accuracy. Studies have been conducted to apply artificial intelligence to shrimp price prediction through automated learning and salmon price volatility through deep learning.

A hybrid approach for Taiwan

This research proposes a novel hybrid model that combines various ML and DL techniques to predict aquatic prices in Taiwan, a region known for its drastic price variations due to various factors, including:

  • Tropical and subtropical climate
  • Fluctuations in fish import/export
  • Uncertainty in political and economic situations

In this context, researchers developed a hybrid approach for Taiwan’s fishing and aquaculture industry, which includes:

Combination of multiple models: This study proposes a hybrid model, integrating six different machine learning and deep learning methods: linear regression, Support Vector Machine (SVM) with various kernels (RBF, linear, and poly), Random Forest, and Long Short-Term Memory (LSTM) network.

  • Data sources: The model utilizes various data, including open fish transactions and meteorological information from the Fisheries Agency, Council of Agriculture (FA.COA), holidays, and news from various sources.
  • Real-time and automated: The model not only predicts future prices but also continuously learns and adjusts to improve accuracy.

The “AIoT SmartFishery” solution

The models work together to predict high, medium, and low prices for eight major types of fish in 14 wholesale markets in Taiwan. Thus, the “AIoT SmartFishery” solution features:

  • Integration with LINE application: This user-friendly application provides real-time and geographically specific fish price information.
  • One-week price trends and predictions: Users can access accurate forecasts with over 90% accuracy.
  • AI Chatbot: This interactive feature provides information on specific fish species and suggests optimal options for consumers and fishermen.
See also  Aquaculture Stewardship Council Pioneers Technology to Verify Seafood Origin With Better Than 95% Accuracy

Innovation beyond predictions

The model goes beyond simple predictions by incorporating:

  • Statistical analysis to optimize input data and predicted outcomes.
  • Regularization techniques to enhance model performance.
  • Self-learning algorithms that automatically acquire new data and adjust model weights for continuous improvement and real-time adaptation.

Benefits for all stakeholders

This innovative model offers significant advantages for all participants in the aquatic ecosystem:

  • Fishermen: Make informed decisions about fishing strategies and maximize profits while adopting sustainable practices.
  • Consumers: Access real-time price information and make informed purchasing decisions.
  • Wholesalers and distributors: Optimize inventory management and business strategies based on accurate price predictions.
  • Government: Utilize scientific insights from the model to formulate effective aquaculture policies.

Conclusion

In conclusion, this innovative fish price prediction model empowers all participants in the aquatic market by harnessing the power of AI. By providing reliable predictions and fostering informed decision-making, it paves the way for a more sustainable, profitable, and secure future for our fisheries.

By enabling informed decision-making across the board, this AI-driven fish price forecasting system paves the way for a more sustainable, profitable, and equitable aquatic market in Taiwan and potentially beyond.

The study was funded by the National Science and Technology Council, Taiwan, and Ming Chi University of Technology.

Contact
Yi-Ting Lai
Department of Materials Engineering, Ming Chi University of Technology, New Taipei City 24301, Taiwan, ROC.
laieating@mail.mcut.edu.tw

Reference
Yi-Ting Lai, Yan-Tsung Peng, Wei-Cheng Lien, Yun-Chiao Cheng, Yi-Ting Lin, Chen-Jie Liao, Yu-Shao Chiu. 2024. Fully automated learning and predict price of aquatic products in Taiwan wholesale markets using multiple machine learning and deep learning methods, Aquaculture, 2024, 740741, ISSN 0044-8486, https://doi.org/10.1016/j.aquaculture.2024.740741.