• 主题
高级搜索  >
历史搜索 清空历史
首页> 外文期刊>Building and environment >Research on risk scorecard of sick building syndrome based on machine learning
【24h】

Research on risk scorecard of sick building syndrome based on machine learning

机译：Research on risk scorecard of sick building syndrome based on machine learning

获取原文
获取原文并翻译 | 示例
           
页面导航
  • 摘要
  • 著录项
  • 相关主题

摘要

Sick building syndrome (SBS) negatively affects numerous aspects of human health. To date, few studies have used machine learning to establish SBS risk prediction models. In this study, we investigated 2370 buildings in Chongqing, China and collected 29,785 questionnaires as the basic dataset. Logistic regression was used as the machine learning algorithm. Based on the weight of evidence and information value, features with significant effects were selected, and prediction models were established for mucosal irritation, dermal symptoms, neuro-toxic symptoms, and respiratory symptoms. Through verification based on the accuracy, receiver operating characteristic curve, and Kolmogorov-Smirnov values, the model performed well on the test set. Thus, the prediction models were further transformed into a scorecard that could provide quantitative results for each environment, and the impact of each level of each feature on the results could be observed. The results indicated that when there was an indoor pollution source, the risks of the four SBS models increased. In addition, the indoor thermal environment had a significant impact on SBS. If the indoor thermal environment deviated from the range of thermal comfort, the prevalence of SBS increased, particularly when the indoor and outdoor thermal environments had opposite thermal sensations to the occupants. Finally, we initially established four SBS rating systems based on quantiles, which can provide a reference for establishing a rating system for healthy indoor environments.

著录项

获取原文

客服邮箱：kefu@zhangqiaokeyan.com

京公网安备：11010802029741号 ICP备案号：京ICP备15016152号-6 六维联合信息科技 (北京) 有限公司©版权所有

  • 客服微信

  • 服务号