Deep Learning for Systemic Food Safety Analytics: Forecasting RASFF Notifications and Interpreting Risk DriversDeep Learning for Systemic Food Safety Analytics: Forecasting RASFF Notifications and Interpreting Risk Driversdatabase_Deep Learning for Systemic Food Safety Analytics: Forecasting RASFF Notifications and Interpreting Risk Drivers.xlsx

This dataset contains the raw monthly RASFF notification counts related to dairy products between 2001 and 2024, along with selected economic, environmental, social, and technological indicators for five EU countries.
The data were compiled to support time-series classification and systemic food safety risk analysis.
Each sheet in the Excel file corresponds to a different indicator category or RASFF alert grouping.
This dataset is intended to accompany the article submitted to Food Control titled "Deep Learning for Systemic Food Safety Analytics: Forecasting RASFF Notifications and Interpreting Risk Drivers".
For academic, non-commercial use only.
This dataset supports the study "Deep Learning for Systemic Food Safety Analytics: Forecasting RASFF Notifications and Interpreting Risk Drivers", which presents an end-to-end deep learning framework to forecast monthly dairy-related food safety notifications from the EU’s Rapid Alert System for Food and Feed (RASFF).

The dataset integrates 15 systemic indicators across economic, environmental, social, and technological domains, covering five EU countries from 2001 to 2024. Indicators include variables such as antibiotic usage, temperature, agricultural income, trade volumes, and other factors influencing systemic food safety risks. 

By combining RASFF notification data with multi-domain systemic indicators, this dataset enables replication of the study’s results and facilitates further research on predictive food safety analytics, early warning systems, and explainable AI (XAI) applications in agri-food contexts. It can be applied in policy development, industry monitoring, and methodological innovation for time-series classification and risk interpretation.

Data type: Time-series, monthly resolution (2001–2024)

File format: [XLSX]