Electronic nose homogeneous data sets for beef quality classification and microbial population prediction.

Area of Science:

• Food Science
• Sensory Analysis
• Data Science

Background:

• Electronic noses (e-nose) coupled with machine learning are advancing food quality assessment.
• Homogeneous datasets are crucial for robust predictive model generalizability.
• Beef quality and microbial spoilage detection are significant food safety concerns.

Purpose of the Study:

• To present novel, homogeneous e-nose datasets for beef quality classification.
• To provide data for microbial population prediction in beef.
• To facilitate the development and validation of machine learning models in food analysis.

Main Methods:

• Utilized 11 Metal-Oxide Semiconductor (MOS) gas sensors to monitor beef spoilage.
• Recorded spoilage process for 12 types of beef cuts over 2220 minutes.
• Classified meat quality into four categories (excellent, good, acceptable, spoiled) based on microbial load and formal standards.

Main Results:

• Generated a comprehensive dataset of beef spoilage indicators using e-nose technology.
• The dataset is formatted in .xlsx files, with each sheet detailing a specific beef cut.
• Data is suitable for testing feature selection algorithms and sensor array optimization.

Conclusions:

• The provided datasets support research in objective beef quality assessment.
• This data facilitates advancements in machine learning applications for food spoilage prediction.
• The homogeneous nature of the data enhances the reliability of developed predictive models.