IoT-Based Bi-Cluster Forecasting Using Automated ML-Model Optimization for COVID-19

The current COVID-19 pandemic has raised huge concerns about outdoor air quality due to the expected lung deterioration. These include challenges associated with an increase of harmful gases like carbon dioxide, iterative/repetitive inhalation mask usage, and harsh environmental temperatures. Even in presence sensing devices, these can hinder prevention treatment respiratory diseases, epidemics, pandemics severe cases. In this research, a dual time series bi-cluster sensor data-stream-based novel optimized regression algorithm was proposed optimization predictors responses that use automated iterative model based on similarity coefficient index. implemented over SeReNoV2 nodes data, i.e., multi-variate time-series sensor, US Environmental Protection Agency standard, which measures variables for index using sensors geospatial profiling. systems were placed at four locations 3 km apart monitor their data collected Ubidots IoT platform GSM. results have shown technique achieved root mean square error (RMSE) 1.0042 training 469.28 s control RMSE 1.646 28.53 when optimized. estimated R-Squared 0.03, Mean-Square Error temperature being 1.0084 °C, 293.98 ppm CO2. Furthermore, Mean-Absolute (MAE) 0.66226 °C 10.252 correlated-CO2 predicted speed ~5100 observations/s. sample cluster temperature, 45,000 observations/s CO2 (469.28 s). correlated very promising forecasting countermeasures before time.