O-ESD
收藏资源简介:
该研究提出了一个高效的工作流程,将建筑性能指南的文本数据转换为适合机器学习的结构化表格数据。此外,结构化楼层布局数据的可视化揭示了分析数据集的新见解。作为本研究的主要产品,Oriented Environmental Swiss Dwellings (O-ESD)数据集为从现有楼层布局数据集到环境设计自动化的数据驱动学习提供了机会。此外,O-ESD通过结构化的微气候可视化提供了人类可解释性。本研究以瑞士住宅数据集为基础,随后进行了基于Python的数据细化、特征工程和属性扩展。修改后的属性包括空间分区(分类)、日光指标和视图层的代理指标(数值)、噪声水平(数值)、声学舒适度(分类)和窗户方向(分类)。
This study proposes an efficient workflow that converts textual data from building performance guidelines into structured tabular data suitable for machine learning. Additionally, visualizations of structured floor layout data uncover new insights for dataset analysis. As the primary product of this research, the Oriented Environmental Swiss Dwellings (O-ESD) dataset provides opportunities for data-driven learning connecting existing floor layout datasets to environmental design automation. Moreover, the O-ESD dataset offers human interpretability through structured microclimate visualizations. This study is grounded in a Swiss residential dataset, followed by Python-based data refinement, feature engineering, and attribute expansion. The revised attributes include spatial zoning (categorical), daylight metrics and proxy indicators for view layers (numerical), noise levels (numerical), acoustic comfort (categorical), and window orientation (categorical).




