Low-cost sensor data for VOC and BTEX calibration

Uncalibrated and calibrated sensor data used in Frischmon, et. al. "Evaluating machine learning model performance in a two-step colocation process for TVOC and BTEX sensor calibration."

The uncalibrated data contains all sensor inputs used to build the calibration models. The calibrated data contains the concentrations predicted from the calibration models.

Please note the data was time-averaged (median) to 15-min intervals prior to publication. The first hour of data collection was also removed prior to publication, as the metal oxide sensors require a warm-up period to reach operating temperatures.

File name key:
{modelname}_colocation_calibrated: The predicted concentrations by a given calibration model during colocation
colocation_raw: The raw secondary standard sensor data collected during colocation
harmonization_secondary-standard: The raw secondary standard sensor data collected during harmonization
harmonization_harmonized: The harmonized sensor data collected during harmonization by field sensors
harmonized_raw: The raw sensor data collected during harmonization by field sensors
{model_name}_field_harmonized-and-calibrated: The predicted field concentrations by a given calibration model
field_raw: The raw sensor data collected in the field

Units:
VOC concentration- ppb
BTEX concentration- ppb
NO2 concentration- ppb

Uncalibrated sensor data does not have units, as it is an uncalibrated sensor signal.