Prediction of traffic noise induced annoyance: A two-staged SEM-Artificial Neural Network approach

The two-staged Structural Equation Modeling-Artificial Neural Network approach was used in this study to assess the annoyance caused by traffic noise in 158 people. The SEM-Partial Least Squares path revealed that sensitivity, exposure hours, profession, sleeping disorder, and education significantly affect annoyance. The variables, such as age, experience, gender, and Leq are found to be inconsequential. The measurement model confirmed 67.5 percent of the variance in annoyance. However, the effectiveness of the Artificial Neural Network model is justified by observing the Mean Square Error and Root Mean Square Error values, and the model's accuracy is 71.2 percent. Furthermore, the feed-forward back-propagation ANN approach confirmed that noise sensitivity is the most important predictor of noise annoyance, followed by exposure hours, profession, sleeping disorder, and education. The SEM-PLS path also revealed that combined socio-demographic factors affect annoyance indirectly through noise sensitivity and sleeping disorder and directly affect annoyance, sensitivity, and sleeping disorder.

Automated summary

This study used the two-staged Structural Equation Modeling-Artificial Neural Network approach to analyze the factors affecting annoyance caused by traffic noise. The results showed that factors such as sensitivity, exposure hours, profession, sleeping disorder, and education significantly influence annoyance levels, while variables like age, experience, gender, and Leq had little impact. The measurement model confirmed 67.5 percent of the variance in annoyance, and the Artificial Neural Network model demonstrated an accuracy of 71.2 percent in predicting annoyance levels. Additionally, socio-demographic factors indirectly affect annoyance through noise sensitivity and sleeping disorder, as well as directly influencing annoyance levels.