Fine-Tuned Extra Tree Classifier for Thermal Comfort Sensation Prediction

Abstract

Thermal comfort is an essential component of smart cities that helps to upgrade, analyze, and realize intelligent buildings. It strongly affects human psychological and physiological levels. Residents of buildings suffer stress because of poor thermal comfort. Buildings frequently use Heating, Ventilation, and Air Conditioning (HVAC) systems for temperature control. Better thermal states directly impact people’s productivity and health. This study revealed a human thermal comfort model that makes better predictions of thermal sensation by identifying essential features and employing a tuned Extra Tree classifier, MultiLayer Perceptron (MLP) and Naive Bayes (NB) models. The study employs the ASHRAE RP-884 standard dataset for experimentation and analysis, which is available to the public. Exploratory Data Analysis (EDA) is performed to examine the outliers and anomalies in the dataset. The Synthetic Minority Over-Sampling Technique (SMOTE) enhances the minority class’s interpretation. The proposed Extra Tree classifier outperforms by achieving an accuracy of 94%. The experiment shows that the suggested model is superior to other established methods and state-of-the-art.