[Paper Review] Understanding occupants' behaviour, engagement, emotion, and comfort indoors with heterogeneous sensors and wearables

Nan Gao, Max Marschall+3 more|arXiv (Cornell University)|May 14, 2021

Building Energy and Comfort Optimization4 citations

TL;DR

This study presents a novel, publicly available dataset combining heterogeneous environmental sensors and wearable physiological sensors (Empatica E4) with self-reported surveys from 29 occupants across a K-12 school in Melbourne, enabling predictive modeling of occupant behavior, emotional states, and thermal comfort. The integration of longitudinal indoor/outdoor climate data with real-time physiological and behavioral metrics offers a foundation for developing intelligent feedback systems to enhance well-being and energy efficiency in educational buildings.

ABSTRACT

We conducted a field study at a K-12 private school in the suburbs of Melbourne, Australia. The data capture contained two elements: First, a 5-month longitudinal field study In-Gauge using two outdoor weather stations, as well as indoor weather stations in 17 classrooms and temperature sensors on the vents of occupant-controlled room air-conditioners; these were collated into individual datasets for each classroom at a 5-minute logging frequency, including additional data on occupant presence. The dataset was used to derive predictive models of how occupants operate room air-conditioning units. Second, we tracked 23 students and 6 teachers in a 4-week cross-sectional study En-Gage, using wearable sensors to log physiological data, as well as daily surveys to query the occupants' thermal comfort, learning engagement, emotions and seating behaviours. Overall, the combined dataset could be used to analyse the relationships between indoor/outdoor climates and students' behaviours/mental states on campus, which provide opportunities for the future design of intelligent feedback systems to benefit both students and staff.

Motivation & Objective

• To address the performance gap in building energy simulations by collecting empirical data on occupant behavior in real-world school environments.
• To investigate the dynamic relationships between indoor climate, physiological responses, and occupant well-being, including emotional states and learning engagement.
• To develop and validate a multi-modal sensing framework using wearable devices and environmental sensors for holistic indoor environment monitoring.
• To create a publicly available, high-fidelity dataset for future research on occupant behavior, comfort, and mental states in educational buildings.
• To enable the design of intelligent feedback systems that support occupant well-being while reducing energy consumption.

Proposed method

• Conducted a 5-month longitudinal field study using outdoor and indoor weather stations, plus temperature sensors on room air-conditioners, logging data at 5-minute intervals with occupant presence detection.
• Carried out a 4-week cross-sectional study using Empatica E4 wristbands to collect physiological signals (EDA, HRV, skin temperature, 3D acceleration) from 23 students and 6 teachers.
• Collected daily self-reported data on thermal comfort, learning engagement, emotions, and seating behavior via digital surveys.
• Applied median filtering with a 5-second window to clean EDA signals and reduce motion artifacts, improving data quality from 2.66% to 2.46% missing values.
• Integrated heterogeneous data streams (environmental, physiological, behavioral, and self-report) into a unified dataset for multi-modal analysis.
• Released the dataset and associated Python code for data preprocessing and segmentation to support reproducibility and further research.

Experimental results

Research questions

• RQ1 How do indoor and outdoor climatic conditions influence the operation of occupant-controlled air-conditioning units in school classrooms?
• RQ2 What is the relationship between physiological signals (e.g., EDA, HRV) and self-reported emotional states or learning engagement among students and teachers?
• RQ3 To what extent can wearable sensor data predict occupant comfort and mental state in real-world educational environments?
• RQ4 How do seating behaviors and peer proximity correlate with physiological synchrony and perceived engagement in classroom settings?
• RQ5 Can integrated sensor data improve the accuracy of occupant behavior modeling in building performance simulations beyond simplistic rule-based assumptions?

Key findings

• The longitudinal environmental dataset, collected at 5-minute intervals across 17 classrooms, enables the development of predictive models for air-conditioner usage based on indoor and outdoor temperature and occupancy.
• The wearable sensor data showed a reduction in missing values from 2.66% to 2.46% after applying a 5-second median filter to EDA signals, enhancing data reliability for physiological analysis.
• The combined dataset provides the first publicly available resource for studying high school students’ daily behaviors, engagement, and emotional states using heterogeneous sensing in real-world school environments.
• The integration of physiological, behavioral, and environmental data opens pathways for detecting disengagement and negative emotions in students, supporting real-time feedback systems.
• The study demonstrates that physiological signals such as EDA and HRV can be effectively used to infer mental states in educational settings when combined with self-report and environmental data.
• The dataset supports future research into peer effects, group-level physiological synchrony, and adaptive indoor climate control to enhance learning and comfort.