Integrating Logis Regression and XGBoost to Construct Indoor Air Quality Improvement Research

Abstract

In the face of the severe global epidemic, indoor architectural space has become one of the critical issues, and the construction of a new type of “built environment” while solving “health and epidemic prevention” has become the goal of active development in countries around the world (SDGs & Pandemic Response); Pollutant concentration, optimization of indoor heat and humidity environment, and release of indoor environmental monitoring data, etc. It can not only protect the short-term needs of building users but also provide long-term health protection for building users and ultimately achieve the purpose of physical and mental health of building users. This study uses GIA-K007-12 Air Box to collect “environmental characteristics” variables; IAQ, PM1, PM2.5, PM10, CO2, TVOC, HCHO, Fungi index, TEMP, and HUMD are input variables for XGBOOST, using IBM SPSS Statistics 20.0 performs statistical analysis, modelling and using PYTHON to simulate the accuracy of the building fresh air system model and the decision ranking of essential factors. The test results are based on the XGBOOST decision tree. The accuracy value reaches 94.24%, and the order of critical environmental factors for the indoor fresh air system is PM1, HCHO, IAQ, Fungi index, TVOC, etc. The research results can provide the basis for constructing a teaching space for epidemic prevention and demonstrate that the establishment of an “air quality control platform that can be calculated in real-time” can improve the environmental health awareness (EHL) of stakeholders and provide for future development of epidemic prevention space planning and design in the post-epidemic era Reference and application of operation management.

Keywords

• decision trees
• indoor air quality
• machine learning
• logis regression