The dataset was developed as part of the EU-funded project Mastermine, designed for training and evaluating machine learning models on time series classification tasks. It comprises 36 classes, each with approximately 3,000 samples—2,400 for training and 300 each for testing and validation. The data is distributed across three CSV files: training.csv, testing.csv, and validation.csv. Each row represents a time series of 100 hourly data points. The final two columns contain the class label and an optional name, generated using large language models. Classes: Cloudiness % CO micrograms/m3 CO ppb CO ppm DewPoint celsius FeelsLike celsius NH3 micrograms/m3 NO2 micrograms/m3 NO2 ppb NO2 ppm NO micrograms/m3 NO ppb NO ppm Ozone micrograms/m3 Ozone ppb Ozone ppm PM10 micrograms/m3 PM2_5 micrograms/m3 Pressure atm Pressure bar Pressure mbar RainRate mm/h RelativeHumidity % SnowRate mm/h SO2 micrograms/m3 SO2 ppb SO2 ppm Temperature celsius Temperature fahrenheit Temperature kelvin Visibility m WindDirection degree WindGust m/s WindSpeed km/h WindSpeed knots WindSpeed m/s Data Sources: Beach Weather Stations - Automated Sensors | City of Chicago | Data Portal Datasets — darts documentation CAIRSENSE-Denver - Catalog Performance Evaluation and Community Application of Low-Cost Sensors for Ozone and Nitrogen Dioxide Data Set - Catalog KC-TRAQS Methods and Measurements - Catalog Smart Green Infrastructure Monitoring Sensors - Historical | City of Chicago | Data Portal Weather API - OpenWeatherMap Download Files | AirData | US EPA