Examination of Two Different Mathematical Techniques for Determining the Important Factors Affecting Indoor Air Quality

摘要

This chapter provides a comprehensive discussion on the performance of regression and regression tree analyses in determining the important factors affecting in-vehicle air quality. Indoor contaminants of particulate matter, carbon dioxide, carbon monoxide, sulfur dioxide, nitric oxide, and nitrogen dioxide were monitored inside two public transport buses. One bus operated on 20% biodiesel and the other operated on ultra-low sulfur diesel in the city of Toledo, Ohio. The independent variables considered in this study include meteorological variables (ambient temperature (temp.), ambient relative humidity (RH), wind speed, wind direction, precipitation, visibility, ambient PM2.5), indoor comfort parameters (indoor temp., indoor RH), and real-time on-road variables (passenger count, bus status (bus position/door position - Idle/Open, Idle/Close, Run/Close), number of cars and buses/trucks ahead). Regression analysis was performed using MINITAB~? software and regression tree analysis was performed using CART~? software. CART proved to have the advantage of providing relative importance of all the independent variables considered. The important factors affecting the monitored vehicular contaminants were found to be different for each month and season on performing the regression and regression tree analyses.