Annoyance from multiple transportation noise: statistical models and outlier detection

Area of Science:

• Environmental Psychology
• Acoustics
• Statistical Modeling

Background:

• Understanding and predicting annoyance from multiple transportation noise sources is crucial for public health and urban planning.
• Existing statistical models require refinement to accurately capture the complex relationship between noise exposure and perceived annoyance.

Purpose of the Study:

• To develop and evaluate statistical models for quantifying annoyance caused by multiple transportation noise exposures.
• To identify the most effective generalized linear model (GLM) and link function for predicting noise-induced annoyance.

Main Methods:

• Generalized linear models (GLMs) were employed to analyze noise annoyance data.
• Robust estimation techniques and outlier detection methods were applied to address data irregularities.
• Various link functions were explored to determine the best fit for the exposure-response relationship.

Main Results:

• A GLM incorporating a complementary log-log link function demonstrated superior performance in modeling the relationship between noise levels and annoyance.
• The inclusion of outlier detection significantly improved the model's explanatory power.
• The chosen model effectively captured the exposure-response curve for transportation noise annoyance.

Conclusions:

• Generalized linear models, particularly with a complementary log-log link and outlier consideration, provide a robust framework for assessing transportation noise annoyance.
• This approach enhances the ability to predict annoyance from specific noise exposures, informing noise mitigation strategies.