Comprehensive Assessment of Indoor Thermal in Vernacular Building Using Machine Learning Model with GAN-Based Data Imputation: A Case of Aceh Region, Indonesia
This study introduces a predictive model for estimating indoor room temperatures in vernacular building using external environmental factors such as air temperature, humidity, sunshine duration, and wind speed. The dataset was sourced from the Meteorology, Climatology, and Geophysics Agency and supp...
Saved in:
| Main Authors: | , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
MDPI AG
2025-07-01
|
| Series: | Buildings |
| Subjects: | |
| Online Access: | https://www.mdpi.com/2075-5309/15/14/2448 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | This study introduces a predictive model for estimating indoor room temperatures in vernacular building using external environmental factors such as air temperature, humidity, sunshine duration, and wind speed. The dataset was sourced from the Meteorology, Climatology, and Geophysics Agency and supplemented with direct measurements collected from four rooms within a vernacular building in Aceh Province, Indonesia. A Generative Adversarial Network (GAN)-based imputation technique was implemented to address missing data during preprocessing. The prediction model adopts a hybrid framework that integrates Multiple Linear Regression (MLR) and Artificial Neural Networks (ANNs), with both models optimized using Support Vector Regression (SVR) to better capture the nonlinear dynamics between inputs and outputs. The evaluation results show that the ANN-SVR model achieved the lowest average <inline-formula><math xmlns="http://www.w3.org/1998/Math/MathML" display="inline"><semantics><mrow><mover><mrow><mi>M</mi><mi>A</mi><mi>E</mi></mrow><mo>¯</mo></mover></mrow></semantics></math></inline-formula> and <inline-formula><math xmlns="http://www.w3.org/1998/Math/MathML" display="inline"><semantics><mrow><mover><mrow><mi>R</mi><mi>M</mi><mi>S</mi><mi>E</mi></mrow><mo>¯</mo></mover></mrow></semantics></math></inline-formula> values, at 0.164 and 0.218, respectively, and the highest average <inline-formula><math xmlns="http://www.w3.org/1998/Math/MathML" display="inline"><semantics><mrow><mover accent="true"><mrow><mi>R</mi></mrow><mo>¯</mo></mover></mrow></semantics></math></inline-formula> and <inline-formula><math xmlns="http://www.w3.org/1998/Math/MathML" display="inline"><semantics><mrow><mover><mrow><msup><mrow><mi>R</mi></mrow><mrow><mn>2</mn></mrow></msup></mrow><mo>¯</mo></mover></mrow></semantics></math></inline-formula> values, at 0.785 and 0.618. Evaluation results indicate that the ANN-SVR model consistently achieved the lowest error rates and the highest correlation coefficients across all four rooms, identifying it as the most effective model for forecasting indoor thermal conditions. These results validate the combined use of ANN-SVR for prediction and GAN for preprocessing as a powerful strategy to enhance data quality and model performance. The findings offer a scientific basis for architectural planning to improve thermal comfort in vernacular buildings such as the Rumoh Aceh. |
|---|---|
| ISSN: | 2075-5309 |