Comparison of Different Classification Algorithms for the Detection of User's Interaction with Windows in Office Buildings

Markovic, Romana; Wolf, Sebastian; Cao, Jun; Spinnräker, Eric; Wölki, Daniel; Frisch, Jérôme; Treeck, Christoph van

doi:10.1016/j.egypro.2017.07.333

Cited by 31 publications

(15 citation statements)

References 18 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…For the RF technique, the need for few and insensitive tuning parameters make it user friendly for parameter optimization. Additionally, the RF algorithm is not prone to overfitting, even for higher characteristic dimensions [54,55]. The most essential ingredients can be selected through RF variable importance functions to construct more concise, readily interpreted, comprehensive, and high-accuracy models.…”

Section: Discussionmentioning

confidence: 99%

Retrieval of Daily PM2.5 Concentrations Using Nonlinear Methods: A Case Study of the Beijing–Tianjin–Hebei Region, China

Chen

Zhang³

et al. 2018

Remote Sensing

View full text Add to dashboard Cite

Exposure to fine particulate matter (PM 2.5 ) is associated with adverse health impacts on the population. Satellite observations and machine learning algorithms have been applied to improve the accuracy of the prediction of PM 2.5 concentrations. In this study, we developed a PM 2.5 retrieval approach using machine-learning methods, based on aerosol products from the Moderate Resolution Imaging Spectroradiometer (MODIS) aboard the NASA Earth Observation System (EOS) Terra and Aqua polar-orbiting satellites, near-ground meteorological variables from the NASA Goddard Earth Observing System (GEOS), and ground-based PM 2.5 observation data. Four models, which are orthogonal regression (OR), regression tree (Rpart), random forests (RF), and support vector machine (SVM), were tested and compared in the Beijing-Tianjin-Hebei (BTH) region of China in 2015. Aerosol products derived from the Terra and Aqua satellite sensors were also compared. The 10-repeat 5-fold cross-validation (10 × 5 CV) method was subsequently used to evaluate the performance of the different aerosol products and the four models. The results show that the performance of the Aqua dataset was better than that of the Terra dataset, and that the RF algorithm has the best predictive performance (Terra: R = 0.77, RMSE = 43.51 µg/m 3 ; Aqua: R = 0.85, RMSE = 33.90 µg/m 3 ). This study shows promise for predicting the spatiotemporal distribution of PM 2.5 using the RF model and Aqua aerosol product with the assistance of PM 2.5 site data.

show abstract

Section: Discussionmentioning

confidence: 99%

Retrieval of Daily PM2.5 Concentrations Using Nonlinear Methods: A Case Study of the Beijing–Tianjin–Hebei Region, China

Chen

Zhang³

et al. 2018

Remote Sensing

View full text Add to dashboard Cite

show abstract

“…Furthermore, we intend to reduce the number of variables needed to run the simulation, e.g. by using predictive window opening models, such as the model developed by Markovic et al, 24 to estimate window opening instead of measuring it, to reach the same accuracies at low number of input variables as the model by Wolf et al 21 A limitation of this work is that we tested the two described models for office rooms exclusively. Testing the applicability for other room types and in particular, the influence of natural ventilation is left to future research.…”

Section: Discussionmentioning

confidence: 99%

Validation, optimisation and comparison of carbon dioxide-based occupancy estimation algorithms

Nienaber

Wolf

Wesseling

et al. 2019

Indoor and Built Environment

Self Cite

View full text Add to dashboard Cite

The operation of heating, cooling and air-conditioning (HVAC) in buildings often adheres to fixed time schedules. However, associating HVAC schedules to the occupant’s presence patterns can save a significant amount of energy, reducing operation periods to the required minimum. Therefore, automated occupancy estimation provides valuable input to efficient building control strategies. This work discusses the validation and adjustment for two carbon dioxide-based occupancy detection algorithms based on data from ten multi-person offices. Both methods are based on a carbon dioxide mass balance equation. However, they follow two different philosophies. One model is deterministic and includes a more detailed representation of the system, whereas the other model includes stochastic elements and was based on fewer assumptions. Both approaches show similar and promising results. The advantages and drawbacks of each method are reviewed. Furthermore, adjustments of the algorithms to the given conditions and possible future improvements are discussed.

show abstract

“…In a study carried out by Markovic et al [11], machine learning classification models are trained to determine the window state during a tenminute time step, given the corresponding values of the input variables (indoor and outdoor climate, occupancy). The trained models can be applied in cases where only the measured variables are known and the goal is to determine the window state for each tenminute time step.…”

Section: Related Researchmentioning

confidence: 99%

Prediction of window handle state using machine learning

et al. 2020

View full text Add to dashboard Cite

Vorhersage des Fensteröffnungszustands unter Anwendung von maschinellem Lernen Der in dieser Publikation beschriebene methodische Ansatz zur Anwendung von Algorithmen des maschinellen Lernens zur Vorhersage des Fensteröffnungsverhaltens von Gebäudenutzenden, basiert auf Monitoring‐Daten eines Bürogebäudes in Regensburg, Deutschland. Das umfangreiche Monitoring zur Erfassung der notwendigen Daten, z. B. der Innenraum‐ und Außentemperaturen, die Fenstergriffzustände etc. wurde im Rahmen eines Forschungsprojektes in den Jahren 2017 bis 2020 durchgeführt. Im Folgenden werden verschiedene Klassifizierungsalgorithmen des maschinellen Lernens auf die gemessenen Daten angewendet, um Modelle für die Vorhersage von Fensteröffnungen und ‐schließungen zu trainieren. Das Ziel ist es, die Potenziale und Grenzen der realistischen Vorhersage des Nutzerverhaltens auf der Grundlage der verfügbaren Daten zu identifizieren.

show abstract

Comparison of Different Classification Algorithms for the Detection of User's Interaction with Windows in Office Buildings

Cited by 31 publications

References 18 publications

Retrieval of Daily PM2.5 Concentrations Using Nonlinear Methods: A Case Study of the Beijing–Tianjin–Hebei Region, China

Retrieval of Daily PM2.5 Concentrations Using Nonlinear Methods: A Case Study of the Beijing–Tianjin–Hebei Region, China

Validation, optimisation and comparison of carbon dioxide-based occupancy estimation algorithms

Prediction of window handle state using machine learning

Contact Info

Product

Resources

About