Predicting Lobby Location as an Indicator of Occupant Comfort: A Machine Learning (ML) Approach

Hospital lobbies function as transitional spaces that affect occupants’ feelings of comfort and efficiency, thereby shaping the overall perception of the healthcare setting. In densely populated cities such as Dhaka, Bangladesh, the interactions between environmental factors (i.e., air quality, illuminance, thermal conditions, and window status), lobby design, and performance have been inadequately explored. This study uses machine learning (ML) to predict lobby location as a metric of perceived occupant satisfaction, recognizing that occupants choose locations based on temperature, visual, and psychological factors. A dataset of 400 samples was collected during the rainy season (June-September) using questionnaires and smart monitoring devices. The dataset contained 24 parameters including demographic detail (i.e., gender, age, visitor type), environmental factors (i.e., CO₂ level, temperature, humidity), spatial attributes (i.e., air conditioning, window status, seating arrangements), and behavioral patterns (i.e., waiting time, crowding). Three machine learning models Decision Tree (DT), Random Forest (RF), and Extreme Gradient Boosting (XGBoost) was applied to predict lobby location. The models’ performance was measured using accuracy, Mean Squared Error (MSE), and Root Mean Squared Error (RMSE); Shapley Additive exPlanation (SHAP) was used for feature ranking. RF achieved the highest prediction accuracy of 96% (MSE = 0.42, and RMSE = 0.65). XGBoost closely followed with an accuracy of 95% (MSE = 0.54, and RMSE = 0.73), while DT was slightly behind with an accuracy of 91% (MSE = 0.76, and RMSE = 0.87). SHAP analysis revealed that Floor_Level and Window_State were the most dominant feature, while Indoor_HCHO and Safety_Guard were the least. This suggesting that vertical positioning and window accessibility are strongly associated with lobby location, whereas Indoor_HCHO and Safety_Guard show minimal association. Thus, the current study provides practical, evidence-based guidance to architects, planners and decision-makers aiming to develop improved occupant centered healthcare settings. Future studies should focus on seasonal variations and a broader range of architectural attributes to design strategies that closely align with occupant needs and can be customized for particular environment.