The dataset of e-nose on beef quality monitoring under an uncontrolled environment

Published: 09-01-2019| Version 1 | DOI: 10.17632/4n23zp92b5.1
Dedy Wijaya,
Enny Zulaika,
Riyanarto Sarno


This dataset is electronic nose signals for beef quality monitoring which labeled for two classes (fresh and spoiled), three classes (fresh, semi-fresh, and spoiled), four classes (excellent, good, acceptable, and spoiled), and continuous labels for regression analysis. In this experiment, the standard of beef quality refers to meat standard issued by Agricultural and Resource Management Council of Australia and New Zealand (CSIRO Food and Nutritional Sciences, 2003). The experiment was performed in the uncontrolled environment using 7 MOS gas sensors. The dataset is divided into training and testing (50%-50%). The explanation of each sheet as follows: Two classes: 1. molen_2class_testing_dwt 2. molen_2class_testing_raw 3. molen_2class_training_dwt 4. molen_2class_training_raw Three classes: 5. molen_3class_testing_dwt 6. molen_3class_testing_raw 7. molen_3class_training_dwt 8. molen_3class_training_raw Four classes: 9. molen_4class_testing_dwt 10. molen_4class_testing_raw 11. molen_4class_training_dwt 12. molen_4class_training_raw Regression: 13. molen_regression_testing_dwt 14. molen_regression_testing_raw 15. molen_regression_training_dwt 16. molen_regression_training_raw “training” and “testing” parts imply data training and data testing, respectively.“The prefix “raw” and “dwt” denote raw and reconstructed signals, respectively. The reconstructed signals use fine-tuned discrete wavelet transform based on Information Quality Ratio (IQR) (Wijaya et al., 2016).