Multisensor Dataset for Alcohol Presence Detection (MQ-3, MQ-135, DHT22)

This dataset has 1,942 rows of sensor data collected to detect alcohol using MQ-3 and MQ-135 gas sensors and a DHT22 sensor for temperature and humidity. Data were gathered in both lab and indoor environments at East West University, Dhaka (23.7833°N, 90.3833°E). The MQ-3 sensor is sensitive to ethanol, while MQ-135 detects various gases. The DHT22 sensor measures the surrounding temperature and humidity. Alcoholic samples had ethanol levels between 5% and 30%, and non-alcoholic samples included water, tea, coffee, and soft drinks. A fixed process was followed for each test. Sensors were preheated for 24 to 48 hours. Each test began with a 10-second waiting time, followed by sensor readings every 30 seconds for 2.5 minutes. A clean air flush was done between samples to clear leftover gas. Each row in the dataset has values from MQ-3, MQ-135, temperature, humidity, and a label showing if alcohol was present (1) or not (0). Tests show that sensor values from MQ-3 and MQ-135 clearly differ when alcohol is present, while temperature and humidity do not show much change. Graphs such as box plots and PCA also show that the two classes are easy to tell apart based on sensor data. The dataset is balanced with 961 alcohol and 981 non-alcohol samples. It is useful for training machine learning models for alcohol detection in cars, workplaces, and portable devices. No people or animals were involved, and data collection followed research ethics.
Please cite the dataset using the following DOI if you use it in your work: 10.17632/kvcfp5y95z.1.