Predicting Fine Spatial Scale Traffic Noise Using Mobile Measurements and Machine Learning

Abstract

Environmental noise has been associated with a variety of health endpoints including cardiovascular disease, sleep disturbance, depression, and psychosocial stress. Most population noise exposure comes from vehicular traffic, which produces fine-scale spatial variability that is difficult to characterize using traditional fixed-site measurement techniques. To address this challenge, we collected A-weighted, equivalent noise (LAeq in decibels, dB) data on hour-long foot journeys around 16 locations throughout Long Beach, California and trained four machine learning models, linear regression, random forest, extreme gradient boosting, and a neural network, to predict noise with 20 m resolution. Input variables to the models included traffic metrics, road network features, meteorological conditions, and land use type. Among all machine learning models, extreme gradient boosting had the best results in validation tests (leave-one-route-out R 2 = 0.71, root mean square error (RMSE) of 4.54 dB; 5-fold R 2 = 0.96, RMSE of 1.8 dB). Local traffic volume was the most important predictor of noise; road features, land use, and meteorology including humidity, temperature, and wind speed also contributed. We show that a novel, on-foot mobile noise measurement method coupled with machine learning approaches enables highly accurate prediction of small-scale spatial patterns in traffic-related noise over a mixed-use urban area.

Applied Machine Learning for Prediction of CO2 Adsorption on Biomass Waste-Derived Porous Carbons

“…19 of the k subsets was used as the validation set and the other k − 1 subsets form the training set to address the variance and bias issue in ML. 30 As a general rule of thumb and depending on the size of the data set, k = 3, 5, or 10 is generally preferred, and much of this also depends on the modeler's intuitions. Specific to this study, the training set had a total of 448 data points (85% of 527).…”

“…XGB, which is a scalable boosting tree and a variant of the GBDTs, employs numerous DTs and uses the weighted quantile search to aid in parallel and distributed computing, resulting in high computational efficiency and prediction accuracy. 30,32 LGB, which is a recent variant of GBDTs, uses gradient-based one-side sampling and exclusive feature bundling to improve computational efficiency without affecting the prediction accuracy of the model. 33 Hyperparameter tuning is the process of finding a set of hyperparameters to attain the optimal model performance.…”

“…During the training process, the k -fold cross-validation ( k -fold CV) method was applied to tune the hyperparameters to simultaneously improve the model’s prediction and generalization capabilities . In k -fold CV, the training data were divided into k subsets such that at each training iteration, one of the k subsets was used as the validation set and the other k – 1 subsets form the training set to address the variance and bias issue in ML . As a general rule of thumb and depending on the size of the data set, k = 3, 5, or 10 is generally preferred, and much of this also depends on the modeler’s intuitions.…”

“…However, in the case of GBDTs, by adding several DTs in a series (e.g., boosting), whereby each subsequent tree minimizes the errors from the previous one, the process of “boosting” becomes highly efficient, thus leading to the high accuracy of GBDT models. XGB, which is a scalable boosting tree and a variant of the GBDTs, employs numerous DTs and uses the weighted quantile search to aid in parallel and distributed computing, resulting in high computational efficiency and prediction accuracy. , LGB, which is a recent variant of GBDTs, uses gradient-based one-side sampling and exclusive feature bundling to improve computational efficiency without affecting the prediction accuracy of the model …”

Evaluation of Acoustic Noise Level and Impulsiveness Inside Vehicles in Different Traffic Conditions

“…However, these models usually use psychoacoustics metrics, focusing on predicting the assessment of the subjective perception of the sound quality of a passenger or driver. Many works contribute to noise prediction using artificial intelligence, providing a model for the traffic noise [53][54][55] and sound quality prediction [56][57][58] contexts.…”