Indoor Environment’s Quality IEQ Forecasting for a Residential Building Using Machine Learning Models

Since we spend 90% of our time indoors, the indoor environment’s quality ‘IEQ’ can either positively or negatively impact people’s health, well-being, and activity. In that sense, predicting IEQ can assist in locating and detecting looming health risks in indoor environments and allow efficient energy use for indoor comfort, enhance the building’s integrated systems’ performances, and mitigate environmental impacts. In the era of digitalization and smart management of buildings, this article proposes a machine learning-based approach to tackle this issue. As such, machine learning can be applied to predict anomalies by identifying and analyzing patterns in data and developing models that are able to forecast target variables. In this way, this paper aims to select suitable machine learning methods in terms of accuracy and root mean square error ‘RMSE’ while considering the function of time to predict IEQ parameters based on 15 regression models, namely Extra Trees Regressor, LGBM Regressor, Gradient Boosting Regressor, K-Neighbors Regressor, Decision Tree Regressor, XGB Regressor, Extra Tree Regressor, Hist. Gradient Boosting Regressor, Random Forest Regressor, Bagging Regressor, Ada Boost Regressor, Gaussian Process Regressor, MLP Regressor, Nu Support Vector Regression, and Support Vector Regression. Finally, the results show that Decision Tree Regressor model serves as the most accurate predictive model for IEQ, with high accuracy in minimum time.