A Gamma GLM Approach to Wildfire Severity Modeling: Washington State Brian Luquin Angel, Elliot Carlsson Project Mentor(s): Yvonne Chueh, PhD Wildfire risk in Washington State has intensified over the past two decades, driven by changing climate conditions, expansion of the wildland-urban interface, and accumulated fuel loads. This paper develops a severity model for wildfire-related property losses in Washington State from 2000 to 2023 using a generalized linear model (GLM) with a Gamma distribution and log link. This is a standard specification for right-skewed, strictly positive loss data. Candidate predictors include slope, mean temperature, precipitation and others. The severity stage uses a Gamma generalized linear model with a log link, predicting expected loss per acre. Presentation Type: Oral Presentation (May 20, 9:30am–5:00pm) Keywords: Gamma GLM, Wildfire Severity Modeling, Washington State, log link, psuedo-absence sampling SOURCE Form ID: 226 Washington State Driver Demographics and Geographic Variation in Crash Severity Selene Rojas Project Mentor(s): Yvonne Chueh, PhD This research investigates the impact of driver demographics and geographic location on crash severity. While the insurance industry traditionally applies higher premiums to policyholders under the age of 25, this study investigates the validity of such risk classifications. Crash data from the Washington State Department of Transportation (WSDOT) spanning 2021 to 2026 are analyzed using multiple regressions with lasso regularization and random forest models to identify key predictors of crash severity. The variables identified for this study are: age (15–24, 25–39, 40-64, and 65+), gender, and location. The Impact of Risky Driving Behaviors on Crash Severity in Washington State Selene Rojas Project Mentor(s): Yvonne Chueh, PhD Washington state has five main road terrains: steep mountain passes, coastal routes, high-desert plateaus, dense urban freeways, and rural backroads. While each terrain presents distinct hazards, human behavior remains a central factor in traffic safety. This study investigates the relationship between risky driving behaviors and crash severity, utilizing motor vehicle accident data from the Washington State Department of Transportation (WSDOT) from 2021 to 2026. A neural network model will be applied to improve the prediction of crash severity based on driver behavior. The key behavioral factors are speeding, alcohol impairment, distracted driving, drowsiness, and seatbelt use. Presentation Type: Oral Presentation (May 20, 9:30am–5:00pm) Keywords: Statistics, crashes, driver injury severity, drivers SOURCE Form ID: 100
Presentation Type: Oral Presentation (May 20, 9:00am–5:00pm) Keywords: Statistics, crashes, driver injury severity, drivers SOURCE Form ID: 120
137
Made with FlippingBook interactive PDF creator