Machine Learning-Based Prediction of Post-Harvest Losses in Egyptian Strawberry Exports: Cold Chain Analytics,
Wael M Elmessery1, Abdallah Elshawadfy Elwakeel2, Eldessoky S Dessoky3
1Agricultural Engineering Department, Faculty of Agriculture, Kafrelsheikh University, Kafrelsheikh, Egypt.
Journal of Food Science
|August 4, 2026
Summary
This study introduces an ML framework to predict post-harvest strawberry losses in Egypt. Key cold chain factors like pre-cooling delay and storage temperature significantly impact export quality and reduce economic losses.
Area of Science:
- Agricultural Economics
- Data Science
- Supply Chain Management
Background:
- Post-harvest losses significantly impact Egypt's strawberry export industry, a global leader in fresh and frozen exports.
- Failures in the cold chain, packaging, and logistics contribute to substantial economic losses.
Purpose of the Study:
- To develop the first machine learning (ML)-based framework for predicting post-harvest losses in Egyptian strawberry exports.
- To benchmark four ML models (Random Forest, XGBoost, SVC, LSTM) for shelf-life estimation, rejection classification, and loss categorization.
- To identify key cold chain predictors using SHapley Additive exPlanations (SHAP) for actionable insights.
Main Methods:
- Developed an ML prediction framework using a synthetic dataset of 800 Egyptian strawberry export lots.
- Benchmarked Random Forest, XGBoost, SVC, and LSTM model architectures.
- Applied SHAP analysis to identify dominant cold chain predictors.
Main Results:
- An ensemble stacking model achieved high accuracy: R²=0.934 for shelf-life, AUC=0.961 for rejection, and Kappa=0.887 for loss category.
- SHAP analysis identified pre-cooling delay, storage temperature, temperature deviations, and transport duration as critical predictors.
- The framework provides actionable guidance for cold chain operators and logistics managers.
Conclusions:
- The developed ML framework effectively predicts post-harvest losses in Egyptian strawberries.
- Optimizing pre-cooling, storage temperature, and transport duration can significantly mitigate economic losses.
- This data-driven approach enhances the efficiency and profitability of the strawberry export supply chain.