UM E-Theses Collection (澳門大學電子學位論文庫)

check Full Text

Prediction analysis about formaldehyde and TVOC in new campus indoor environment

English Abstract

Prediction Analysis about Formaldehyde and TVOC in New Campus Indoor Environment by Su Peng Thesis Supervisor: Prof. Simon Fong M.Sc. in E-Commerce Technology Along with social development and people's living standards improving indoor air quality (IAQ) is mentioned more frequently. The total volatile organic compounds (TVOC) and Formaldehyde (FA) are major indoor air pollutants which are especially present in densely populated indoor areas, such as school dormitory, residential and office buildings. World Health Organization (WHO) offers safety standards and some institutions also provide air quality monitoring and testing services. But there have not any fast and accurate approach to such a large scale testing as new campus plans of the University of Macau. Therefore, we are measuring buildings at the Macao University while establishing database for machine learning and data mining to predict the result. At the same time, we will adopt a null hypothesis analysis and regression analysis methods to validate some subjective speculation about pollutants, such as floors, weather and seasonal factors impact on pollutions. Passive sampling method was used for measuring the formaldehyde concentration and TVOC in selected rooms which were matched the pre-determined criteria. These rooms are set four in a group and scattered throughout every layers of every buildings on campus. The selection of measurement methods, depends on detection purposes and sample features, but also suitable for measuring requirements. Specific steps are described in detail in articles. To restore the authenticity of the measurement results for the guidelines, did not remove the outliner in the results, but uses a discretization of continuous attributes, using properties of the classification dealing with fuzzy parameters in order to ensure meeting the software requirements for data formats. In four different classification algorithms, the multilayer perception algorithm won the highest degree of accuracy at 93.45%. In five different predictions, the prediction accuracy of formaldehyde concentration higher than the prediction accuracy of TVOC, describes these indicators much closer to the influencing factors of formaldehyde. The two level classification better than three level classification so that the rationality of classification method will affect the accuracy of forecasts. The result of null hypothesis illustrated that there has no significant difference in formaldehyde concentration and TVOC between the different floors mean-vectors. But the different weathers and seasons will impact prediction results. The regression analysis shows that the formaldehyde concentration and TVOC in same room has significant correlation, so they will influence each other. In the future work, it will be meaningful attempt to explore other models to describe the prediction of pollutants in densely populated areas like University of Macau.

Issue date



Su, Peng


Faculty of Science and Technology


Department of Computer and Information Science





Volatile organic compounds -- Analysis

Volatile organic compounds -- Macau

Indoor air quality -- Macau


Fong, Chi Chiu

Files In This Item

Full-text (Internet)

1/F Zone C
Library URL