Datasets
Standard Dataset
Electronic nose data set for seafood quality assessment
- Citation Author(s):
- Submitted by:
- Dedy Wijaya
- Last updated:
- Sat, 05/20/2023 - 22:36
- DOI:
- 10.21227/mddf-qn59
- Data Format:
- License:
- Categories:
- Keywords:
Abstract
The e-nose device used in this study was constructed using a gas sensor array, LCD display, micro air pumps for inhalation and exhalation, a microcontroller, and a mini-PC. Gas samples from the sample chamber were periodically drawn into the device through a hose. Each sample underwent a 30-hour sampling process at room temperature (25°C). The sampling frequency was 15 times per hour, resulting in 60 records per sample. Overall, the experiment yielded a dataset comprising 108,000 records (4 samples × 30 hours × 15 sampling/hour × 60 records). To categorize the recorded data, it was labeled based on quality and microbial population, with reference to established standards. The growth of microbial populations was determined using the Food Spoilage and Safety Predictor (FSSP), a software tool designed to predict the growth of spoilage and pathogenic microorganisms in food. The dataset includes seven features related to the gas sensors used (MQ136, MQ137, MQ5, MQ8) and two labels: TVC and Label. TVC represents the continuous label of microbial population and is calculated using the FSSP calculation from log (cfu/g) values. The Label indicates the discrete label of seafood quality, classified as "accept" or "reject." Seafood quality is considered "accepted" when the log (cfu/g) value is below 4.3, a threshold determined based on laboratory tests conducted on rejected samples.
Comments
good