Applying a Multilinear Regression Model to Predict Air Quality in Burns, Oregon

Author(s)
Travis
Lowe
*,
Eastern Oregon University
Sydney
Nelson
*,
Eastern Oregon University
Amy
Yielding
,
Eastern Oregon University
Talk Abstract
The City of Burns, Oregon has a serious air quality issue. The city frequently experiences very high levels of PM2.5. PM2.5 consists of a variety of particulates whose size is less than 2.5 micrometers. Such particulates can be inhaled and generally accumulate in the lungs of humans, displaying a strong positive correlation with the instance of lung cancers. Field burning and wood burning for heat combined with the unique atmospheric conditions in Burns appear to be contributing to these high levels. In this talk we discuss the methods used in establishing a regression model to predict PM2.5 for Burns. In particular, we worked in collaboration with The National Oceanic and Atmospheric Administration (NOAA) as well as The Oregon Department of Environmental Quality (ODEQ), who had established a preliminary regression equation. Improvements we made to this model were adding interaction terms and including data gathered from the previous day. The resulting equation improved the predictions of PM2.5 by a 10% higher R-squared adjusted and displayed a narrower range of errors. Our model is now implemented as part of a pilot air quality alert system for the City of Burns sent out by ODEQ in collaboration with NOAA.
Talk Subject
Probability and Statistics
Time Slot
2016-04-02T11:15:00
Room Number
STAG 263