Validation of EEATC a Novel Calibration Approach for Low-cost Sensors in Different Tandem Configurations

Low-cost sensors (LCS) for air quality monitoring have gained significant popularity due to their potential to provide real-time data and insights about air pollution in various environments. These sensors work on affordable technologies and offer air quality measurements, which can be used to supplement the air quality data provided by regulatory agencies. Calibration is a known technique to make measurements of LCS more reliable. Estimated Error Augmented Two-phase Calibration (EEATC), is a recently developed novel approach, where two models are cascaded through the estimated errors to improve the calibration accuracy of low-cost sensors. EEATC can calibrate LCS in stationary and mobile deployments. However, it requires empirical validations of the order of models in which they are cascaded. This work is the extension of EEATC approach, and it validates the calibration performance of different tandem configurations of EEATC by changing the order of the cascaded models. We show that EEATC: linear to linear configuration, where a linear model cascaded to a linear model, outperforms the other configurations of EEATC, multiple linear regression (MLR) and random forest (RF) models for different LCS in the CAIRSENSE data set. It shows an 80.1 % and 26.6 % improvement in R
²
(coefficient of determination) compared to the MLR and RF models, respectively, for Speck Sensors.