Real time forecasting of indoor CO2 concentration using random forest.

Nowadays, the presence of humans significantly impacts indoor air quality, necessitating continuous monitoring of pollutant gases, especially CO2, for a healthy living environment. The proposed framework leverages machine learning, specifically the Random Forest algorithm, known for its versatility and accuracy, to predict CO2 levels in real-time. By optimizing the window size through extensive experimentation, the framework achieves the lowest Root Mean Squared Error (RMSE) of 12.0045 at 13. On the other hand, Mean Absolute Percentage Error (MAPE) analysis affirm the framework's high accuracy, consistently maintaining a percentage error below 10%. [ABSTRACT FROM AUTHOR]