As of 2015, there are over 60 million smart meters installed in the United States; these meters are at the forefront of big data analytics in the building industry. However, only a few public data sources of hourly non-residential meter data exist for the purpose of testing algorithms. This paper describes the collection, cleaning, and compilation of several such data sets found publicly on-line, in addition to several collected by the authors. There are 507 whole building electrical meters in this collection, and a majority are from buildings on university campuses. This group serves as a primary repository of open, non-residential data sources that can be built upon by other researchers. An overview of the data sources, subset selection criteria, and details of access to the repository are included. Future uses include the application of new, proposed prediction and classification models to compare performance to previously generated techniques.

This study focuses on the inference of characteristic data from a data set of 507 non-residential buildings. A two-step framework is presented that extracts statistical, model-based, and pattern-based behavior. The goal of the framework is to reduce the expert intervention needed to utilize measured raw data in order to infer information such as building use type, performance class, and operational behavior. The first step is temporal feature extraction, which utilizes a library of data mining techniques to filter various phenomenon from the raw data. This step transforms quantitative raw data into qualitative categories that are presented in heat map visualizations for interpretation. In the second step, a random forest classification model is tested for accuracy in predicting primary space use, magnitude of energy consumption, and type of operational strategy using the generated features. The results show that predictions with these methods are 45.6% more accurate for primary building use type, 24.3% more accurate for performance class, and 63.6% more accurate for building operations type as compared to baselines.

This paper discusses the creation of targeting and segmentation information about non-residential buildings that are equipped with advanced metering infrastructure (AMI) meters, or smart meters. Statistics, model, and pattern-based temporal features are extracted from over 36,000 smart meters. They are then merged with a database of past energy efficiency interventions such as lighting, HVAC, and controls retrof its from 1,600 buildings. The buildings are divided into Good, Average, and Poor performing classes according to consumption from before and after the retrofits. Classification models are developed that improve the ability to predict retrofit success and standard industry class by 18.3% and 27.6% respectively over baselines. This study serves as an example of better leveraging smart meter data from non-residential buildings for utility targeted incentive programs. The methodology outlined is preliminary and further models and temporal features are to be tested.

Building energy performance benchmarking has been adopted widely in the USA and Canada through the Energy Star Portfolio Manager platform. Building operations and energy management professionals have long used this simple 1–100 score to understand how their building compares to its peers. This single number is easy to use but is created by potentially inaccurate multiple linear regression (MLR) models and lacks much further information about why a building achieves that score. This paper proposes a methodology that enhances the existing Energy Star calculation method by increasing accuracy and providing additional model output processing to help explain why a building is achieving a particular score. Two new prediction models were proposed and tested: multiple linear regression with feature interactions (MLRi) and gradient boosted trees (GBT). Both models performed better than a baseline Energy Star MLR model as well as four baseline models from previous benchmarking studies. This paper shows that for six building types, on average, the third-order MLRi models achieved a 4.9% increase in adjusted R2 and a 7.0% decrease in normalized root mean squared error (NRMSE) over the baseline MLR model. More substantially, the most accurate GBT models, on average, achieved a 24.9% increase in adjusted R2 and a 13.7% decrease in NMRSE against the baseline MLR model. In addition, a set of techniques was developed to help determine which factors most influence a building's energy use versus its peers using SHapley Additive exPlanation (SHAP) values. The SHAP force visualization, in particular, offered an accessible overview of the aspects of the building that influenced the score that even non-technical users can interpret. This methodology was tested on the 2012 Commercial Building Energy Consumption Survey (CBECS)(1,812 buildings) and public data sets from the energy disclosure programs of New York City (11,131 buildings) and Seattle (2,073 buildings).

Thermal comfort assessment for the built environment has become more available to analysts and researchers due to the proliferation of sensors and subjective feedback methods. These data can be used for modeling comfort behavior to support design and operations towards energy efficiency and well-being. By nature, occupant subjective feedback is imbalanced as indoor conditions are designed for comfort, and responses indicating otherwise are less common. This situation creates a scenario for the machine learning workflow where class balancing as a pre-processing step might be valuable for developing predictive thermal comfort classification models with high-performance. This paper investigates the various thermal comfort dataset class balancing techniques from the literature and proposes a modified conditional Generative Adversarial Network (GAN), comfortGAN, to address this imbalance scenario. These approaches are applied to three publicly available datasets, ranging from 30 and 67 participants to a global collection of thermal comfort datasets, with 1,474; 2,067; and 66,397 data points, respectively. This work finds that a classification model trained on a balanced dataset, comprised of real and generated samples from comfortGAN, has higher performance (increase between 4% and 17% in classification accuracy) than other augmentation methods tested. However, when classes representing discomfort are merged and reduced to three, better imbalanced performance is expected, and the additional increase in performance by comfortGAN shrinks to 1 - 2%. These results illustrate that class balancing for thermal comfort modeling is beneficial using advanced techniques such as GANs, but its value is diminished in certain scenarios. A discussion is provided to assist potential users in determining which scenarios this process is useful and which method works best.

Personal thermal comfort models are a paradigm shift in predicting how building occupants perceive their thermal environment. Previous work has critical limitations related to the length of the data collected and the diversity of spaces. This paper outlines a longitudinal field study comprising 20 participants who answered Right-Here-Right-Now surveys using a smartwatch for 180 days. We collected more than 1080 field-based surveys per participant. Surveys were matched with environmental and physiological measured variables collected indoors in their homes and offices. We then trained and tested seven machine learning models per participant to predict their thermal preferences. Participants indicated 58% of the time to want no change in their thermal environment despite completing 75% of these surveys at temperatures higher than 26.6°C. All but one personal comfort model had a median prediction accuracy of 0.78 (F1-score). Skin, indoor, near body temperatures, and heart rate were the most valuable variables for accurate prediction. We found that ≈250-300 data points per participant were needed for accurate prediction. We, however, identified strategies to significantly reduce this number. Our study provides quantitative evidence on how to improve the accuracy of personal comfort models, prove the benefits of using wearable devices to predict thermal preference, and validate results from previous studies.

Cohort Comfort Models (CCM) are introduced as a technique for creating a personalized thermal prediction for a new building occupant without the need to collect large amounts of individual comfort-related data. This approach leverages historical data collected from a sample population, who have some underlying preference similarity to the new occupant. The method uses background information such as physical and demographic characteristics and one-time onboarding surveys (satisfaction with life scale, highly sensitive person scale, personality traits) from the new occupant, as well as physiological and environmental sensor measurements paired with a few thermal preference responses. The framework was implemented using two personal comfort datasets containing longitudinal data from 55 people. The datasets comprise more than 6000 unique right-here-right-now thermal comfort surveys. The results show that a CCM that uses only the one-time onboarding survey information of an individual occupant has generally as good or better performance as compared to conventional general-purpose models, but uses no historical longitudinal data as compared to personalized models. If up to ten historical personal preference data points are used, CCM increased the thermal preference prediction by 8% on average and up to 36% for half of the occupants in the first of the tested datasets. In the second dataset, one-third of the occupants increased their thermal preference prediction by 5% on average and up to 46%. CCM can be an important step toward the development of personalized thermal comfort models without the need to collect a large number of datapoints per person.

Global climate change is a clear and present danger to our environment, but the impacts of climate change on human health are less known. People in Asian countries are more susceptible to the negative impacts of climate change and the subsequent environmental exposures because of the high population density, rapid urbanization, and natural geography of the region. The objective of this multidisciplinary collaborative ecological study was to explore the impact of environmental exposures such as temperature (°C), noise (db), humidity (%rh), air conditioning exposure time (hours), and distance traveled to school (km) on the comfort and academic success of school children in Singapore. Analysis of a large dataset from the Singapore National Science Experiment revealed a positive correlation between the distance traveled to school and favorable environmental conditions (moderate temperatures, low noise, low humidity, and higher amount of air conditioning time) and student academic performance. The analysis revealed that the distance traveled between home and school for public school students falls within a larger range than that for independent (private) school students. On average, students traveled farther distances to attend schools of higher academic caliber thereby increasing their exposure to environmental pollution. Student exposure to pollution can be minimized if all schools adhere to higher standards of environmental comfort and standardized academic caliber. If students can attend the school closest to their homes, they can minimize their daily pollution exposure due to traffic/commute, thereby mitigating the resultant negative health consequences.

Collecting feedback from people in indoor and outdoor environments is traditionally challenging and complex to achieve in a reliable, longitudinal, and non-intrusive way. This paper introduces Cozie Apple, an open-source mobile and smartwatch application for iOS devices. This platform allows people to complete a watch-based micro-survey and provide real-time feedback about environmental conditions via their Apple Watch. It leverages the inbuilt sensors of the smartwatch to collect physiological (e.g., heart rate, activity) and environmental (sound level) data. This paper outlines data collected from 48 research participants who used the platform to report perceptions of urban-scale environmental comfort (noise and thermal) and contextual factors such as who they were with and what activity they were doing. The results of 2,400 micro-surveys across various urban settings are illustrated in this paper, showing the variability of noise-related distractions, thermal comfort, and associated context. The results show that participants experienced at least a little noise distraction 58% of the time, with people talking being the most common reason (46%). This effort is novel due to its focus on spatial and temporal scalability and the collection of noise, distraction, and associated contextual information. These data set the stage for larger deployments, deeper analysis, and more helpful prediction models toward better understanding the occupants’ needs and perceptions. These innovations could result in real-time control signals to building systems or nudges for people to change their behavior.