Electronic nose dataset for detection of wine spoilage thresholds

This dataset is related to the research paper "Wine quality rapid detection using a compact electronic nose system: application focused on spoilage thresholds by acetic acid" published in LWT journal (available online from April 1, 2019, https://doi.org/10.1016/j.lwt.2019.03.074), and the data paper "Electronic nose dataset for detection of wine spoilage thresholds" submitted to Data in Brief journal. For more details read the mentioned articles and cite our work whether found useful.

The recorded time series was acquired at the sampling frequency of 18.5Hz during 180 seconds, resulting in 3330 data points per sensor.

Each file in the dataset has eight columns: relative humidity (%), temperature (°C), and the resistance readings in kΩ of the six gas sensors: MQ-3, MQ-4, MQ-6, MQ-3, MQ-4, MQ-6.

We organized the database in three folders for the wines: AQ_Wines, HQ_Wines, LQ_Wines; and one folder for the ethanol: Ethanol. Each folder contains text files that correspond to different measurements.

The filename identify the wine measurement as follows: the first 2 characters of the filename are an identifier of the spoilage wine threshold (AQ: average-quality, HQ: high-quality, LQ: low-quality); characters 4-9 indicate the wine brand; characters 11-13 indicate the bottle, and the last 3 characters indicate the repetition (another sample of the same bottle). For example, file LQ_Wine01-B01_R01 contains the time series recorded when low-quality wine of the brand 01, bottle 01, sample 01 was measured.

The filenames into the Ethanol folder identify the measurements at different concentrations: the first 2 characters of the filename are an identifier of Ethanol (Ea); characters 4-5 indicate the concentration in v/v (C1: 1%, C2: 2.5%, C3: 5%, C4: 10%, C5: 15%, C6: 20%); and the last 3 characters indicate the repetition. For example, file Ea-C1_R01 contains time series acquired when Ethanol at 1% v/v of concentration, sample 01 was measured.