A platform for research: civil engineering, architecture and urbanism
A platform for research: civil engineering, architecture and urbanism
Using random forests to predict passengers’ thermal comfort in underground train carriages
https://orcid.org/0000-0003-0210-8894
This research developed an intelligent ensemble machine learning prediction model for the thermal comfort of passengers inside the compartment of the subway. Data sources used for data-driven modelling were obtained from on-site measurements and passengers’ questionnaires in the compartments of the Nanjing subway. The four models were established using methodologies of Logistic Regression (LR), Random Forest (RF), Support Vector Machine (SVM) and Decision Tree (DT) in machine learning, respectively. The performance of the RF method was compared with DT, LR and SVM in terms of conventional statistical metrics, namely, Mean Squared Error (MSE), Root Mean Square Error (RMSE) and Correlation Coefficients squares (R2). Thermal Sensation Vote with the seven-level indicator (TSV-7) and Thermal Sensation Vote with the three-level indicator (TSV-3) were employed to obtain passengers’ thermal comfort and evaluate the models’ predictions. In this study, the R2 value of the RF model is 0.6527 and 0.6607 for TSV-7 and TSV-3, which shows higher accuracy than DT, LR and SVM models in predicting the two kinds of Thermal Sensation Vote (TSV). The results show that the predictive performance of the proposed RF model is outstanding, and it can predict the TSV value of passengers inside the compartment of the subway more efficiently.
Using random forests to predict passengers’ thermal comfort in underground train carriages
https://orcid.org/0000-0003-0210-8894
This research developed an intelligent ensemble machine learning prediction model for the thermal comfort of passengers inside the compartment of the subway. Data sources used for data-driven modelling were obtained from on-site measurements and passengers’ questionnaires in the compartments of the Nanjing subway. The four models were established using methodologies of Logistic Regression (LR), Random Forest (RF), Support Vector Machine (SVM) and Decision Tree (DT) in machine learning, respectively. The performance of the RF method was compared with DT, LR and SVM in terms of conventional statistical metrics, namely, Mean Squared Error (MSE), Root Mean Square Error (RMSE) and Correlation Coefficients squares (R2). Thermal Sensation Vote with the seven-level indicator (TSV-7) and Thermal Sensation Vote with the three-level indicator (TSV-3) were employed to obtain passengers’ thermal comfort and evaluate the models’ predictions. In this study, the R2 value of the RF model is 0.6527 and 0.6607 for TSV-7 and TSV-3, which shows higher accuracy than DT, LR and SVM models in predicting the two kinds of Thermal Sensation Vote (TSV). The results show that the predictive performance of the proposed RF model is outstanding, and it can predict the TSV value of passengers inside the compartment of the subway more efficiently.
Using random forests to predict passengers’ thermal comfort in underground train carriages
https://orcid.org/0000-0003-0210-8894
This research developed an intelligent ensemble machine learning prediction model for the thermal comfort of passengers inside the compartment of the subway. Data sources used for data-driven modelling were obtained from on-site measurements and passengers’ questionnaires in the compartments of the Nanjing subway. The four models were established using methodologies of Logistic Regression (LR), Random Forest (RF), Support Vector Machine (SVM) and Decision Tree (DT) in machine learning, respectively. The performance of the RF method was compared with DT, LR and SVM in terms of conventional statistical metrics, namely, Mean Squared Error (MSE), Root Mean Square Error (RMSE) and Correlation Coefficients squares (R2). Thermal Sensation Vote with the seven-level indicator (TSV-7) and Thermal Sensation Vote with the three-level indicator (TSV-3) were employed to obtain passengers’ thermal comfort and evaluate the models’ predictions. In this study, the R2 value of the RF model is 0.6527 and 0.6607 for TSV-7 and TSV-3, which shows higher accuracy than DT, LR and SVM models in predicting the two kinds of Thermal Sensation Vote (TSV). The results show that the predictive performance of the proposed RF model is outstanding, and it can predict the TSV value of passengers inside the compartment of the subway more efficiently.
Using random forests to predict passengers’ thermal comfort in underground train carriages
Huang, Kangkang (author) / Lu, Shihua (author) / Li, Xinjun (author) / Chen, Weiwei (author)
Indoor and Built Environment ; 32 ; 343-354
2023-02-01
12 pages
Article (Journal)
Electronic Resource
English
| SAGE Publications | 2024
Characteristics of fire in high-speed train carriages
| SAGE Publications | 2020
Train delay and perceived-wait time: passengers' perspective
| Taylor & Francis Verlag | 2014
Thermal comfort of novel low-momentum ventilation concepts for passengers in long-distance trains
| SAGE Publications | 2023
Indoor environment and passengers’ comfort in subway stations in Seoul
| Online Contents | 2016