An integrated model for facilities management: Indoor environment evaluation

Several definitions of the sick building have been investigated. A blended definition of a closely defined sick building, heavily involving the building occupant's perceptions was then summarized to begin this research. The management of the sick building by facility managers was investigated. A statistical analysis of factors causing the occupant's perception of sick buildings was conducted, and it was concluded that the factors are conditional and thus not suitable for a regression analysis. An exploratory data analysis of the significant variables causing building occupants to perceive a sick building was then conducted. A decision-making support model for the facility manager to proactively manage the occupant's perception of sick buildings was developed. This dissertation reports the development of the "proactive remediation of environmental detractors indoors, utilizing carbon dioxide measurements and occupant thresholds" model "PREDICT". PREDICT embraces the many-faceted nature of the sick building, and balances sick building management by including the current carbon dioxide measurement technology with occupant perception metrics in a neural network based model. Several different combinations of factors were employed to train 17 neural networks that were then tested for accuracy of prediction. The best performing network was chosen as the basis for the PREDICT model.