Salmon Consumption Behavior Prediction Based on Bayesian Optimization and Explainable Artificial Intelligence.

Area of Science:

• Agricultural Economics
• Consumer Behavior Analysis
• Machine Learning Applications

Background:

• Accurate prediction of seafood consumption is crucial for optimizing production and marketing strategies in the fishing industry.
• Understanding consumer purchasing intentions, particularly for high-demand items like salmon, is essential for market success.

Purpose of the Study:

• To develop and evaluate an interpretable machine learning model for forecasting seafood consumption behavior.
• To identify and analyze the key factors influencing Shanghai residents' intentions to purchase salmon.
• To compare the performance of various regression models and optimize the best-performing one.

Main Methods:

• Construction and comparison of nine regression prediction models (ANN, Decision Tree, GBDT, Random Forest, AdaBoost, XGBoost, LightGBM, CatBoost, NGBoost).
• Integration of Bayesian optimization for hyperparameter tuning of the optimal model.
• Application of Shapley Additive exPlanations (SHAP) and Accumulated Local Effects (ALE) plots for factor analysis.

Main Results:

• The Bayesian optimization-tuned CatBoost (BO-CatBoost) nonlinear regression model significantly outperformed benchmark models.
• Salmon farming safety and ease of cooking were identified as significant nonlinear factors influencing salmon consumption.
• The BO-CatBoost model achieved high predictive accuracy with specific performance metrics (RMSE, MSE, MAE, R², TIC).

Conclusions:

• The developed BO-CatBoost model provides a robust tool for predicting seafood consumption behavior.
• Insights into consumer preferences, particularly regarding salmon attributes, can guide strategic business decisions.
• This research offers valuable technical support for salmon value chain stakeholders to enhance production and marketing efforts.