Humans spend approximately 90% of the daytime in buildings, and greenhouse gases (GHGs) emitted by buildings account for approximately 20% of total GHG emissions. As the energy consumed during building operation from a building life-cycle perspective amounts to approximately 70–90% of the total energy, it is essential to accurately predict the energy consumption of buildings for their efficient operation. This study aims to optimize a model for predicting the thermal energy consumption of buildings by (i) first extracting major variables through feature selection and deriving significant variables in addition to the collected data and (ii) predicting the thermal energy consumption using a machine learning model. Feature selection using random forest was performed, and 11 out of 17 available data were selected. The accuracy of the prediction model was significantly improved when the hour of day variable was added. The prediction model was constructed using an artificial neural network (ANN), and the improvement in the prediction accuracy was analyzed by comparing different cases of variable combinations. The ANN prediction accuracy was improved by 15% using the feature selection process compared to when all data were used as input data, and 25% coefficient of variation of the root mean square error (CVRMSE) accuracy was achieved.

It has been identified that improving building energy efficiency is an effective method to reduce greenhouse gas (GHG) emissions. Although standards have been established to satisfy a building’s minimum energy demand while ensuring the comfort of its residents, they are difficult to implement in mixed-humid regions. This study proposes a hybrid ventilation strategy that can comprehensively reduce cooling, heating, and ventilation energy in mixed-humid climate regions to significantly decrease the primary energy demand and reduce the impact of buildings on the environment. This study evaluated the changes in energy saving potential and thermal comfort according to the extension of the natural ventilation period and passive strategies, such as decentralized ventilation. Changes in indoor air temperature, operative temperature, and PMV for each strategy were analyzed. As a result, extending the natural ventilation and the decentralized ventilation strategies can save 32% and 34% of the building’s energy, respectively. Considering that electricity is the main energy source for cooling in Korea, the extension of the natural ventilation period was judged to be the best approach from the perspective of primary energy demand. The results can be used to predict changes in building energy demand and thermal comfort and select an appropriate ventilation strategy based on occupant information obtained using Internet of Things.

Since P.O. Fanger proposed PMV, it has been the most widely used index to estimate thermal comfort. However, in some cases, it is challenging to measure all six parameters within indoor spaces, which are essential for PMV estimation; a couple of parameters, such as Clo or Met, tend to show a large deviation in accuracy. For these reasons, several studies have suggested methods to estimate PMV but their accuracies were significantly compromised. In this vein, this study proposed a way to reduce the dimensions of parameters for PMV prediction utilizing the machine learning method, in order to provide fast PMV calculations without compromising its prediction accuracy. Throughout this study, the most influential features for PMV were pinpointed using PCA, Best Subset, and the Gini Importance, with each model compared to the others. The results showed that PCA and ANN achieved the highest accuracy of 89.70%, and the combination of Best Subset and Random Forest showed the fastest prediction performance among all.