Aquaculture for all
Introducing our new subscription model Get a free account today to lock in your 25% future discount

Using machine learning to predict shrimp prices

Shrimp Technology & equipment Economics +4 more

Researchers from Can Tho University and Hokkaido University have used machine learning to predict export price trends for Vietnamese shrimp, based on information from the world’s top shrimp exporting countries.

The study provides a new approach to making reliable predictions about Vietnamese shrimp prices, based on competitor data rather than just Vietnamese data.

The application of computerised techniques to improve management, predict diseases or analyse market trends is widely used. For example, an expert system and digital image processing tools were used to diagnose shrimp diseases. Although machine learning algorithms are useful for making predictions, they depend on a reliable database.

Price prediction is important for fishery and aquaculture exports because it helps determine global market trends and increases the quality of seafood products.

A shrimp farmer in Vietnam

Vietnam is one of the top seven shrimp exporters in the world © ASC

Databases used

The researchers used the databases of the US Department of Agriculture (USDA), the Food and Agriculture Organization of the United Nations (FAO), and the World Trade Organization (WTO), in the period from May 1995 to May 2019.

The seven leading exporters of frozen shrimp products for the US market – China, Thailand, India, Indonesia, Ecuador, Chile and Vietnam – were included in the database.

The exporting countries were direct competitors in the US market; therefore, an increase in the export price of any of the listed countries may change the demand curve for shrimp products imported into the US market.

A technique known as “super learner”, which combines 10 simple algorithms, was used to make the predictions in selected base periods (3, 6, 9, and 12 months).

To interpret the prediction, the researchers used the SHapley Additive exPlanations (SHAP) method to determine how each predictor influences the export price and then suggest solutions to develop the Vietnamese shrimp industry.

The researchers found that the super learner returned results across all base periods that were more approximate and stable than any other candidate algorithm.

Graphs showing how machine learning driven price predictions (red lines) measure up the the real prices (black lines)

Predictions by the “super learner” of shrimp prices (a) 3 months, (b) six months, (c) 9 months and (d) 12 months. © Khiem et al, 2022

According to the study, the SHAP analysis results highlight that the price difference with India, WHO membership, the price difference with Thailand, the price difference with China, the early mortality syndrome (EMS) outbreak and the price difference price with Ecuador had the greatest impacts on the prediction of Vietnamese shrimp prices.

“Although other factors, including Aquaculture Steward Council, Global GAP, Safe Quality Food, and HACCP certificates, had less impact on the predictive model, they also impact the export price of Vietnam and other countries in terms of ensuring the health, traceability, and risk of disease,” the researchers noted.

According to the study, Vietnam is likely to get better prices for its shrimp products if it fully implements shrimp export safety certificates, which can give it a competitive advantage over other producing countries in the international market.

“The prices of exports from India, Thailand, and China highlight the advantage of being members of the World Trade Organization (WTO) and the disadvantage of the prevalence of shrimp diseases in Vietnam, which had a significant impact on the export price of Vietnamese shrimp,” they conclude.

Reference (open access)

Khiem NM, Takahashi Y, Yasuma H, Dong KTP, Hai TN, Kimura N (2022) A novel machine learning approach to predict the export price of seafood products based on competitive information: The case of the export of Vietnamese shrimp to the US market. PLoS ONE 17(9): e0275290. https://doi.org/10.1371/journa...

Create an account now to keep reading

It'll only take a second and we'll take you right back to what you were reading. The best part? It's free.

Already have an account? Sign in here