Analysis of relationship between road traffic noise and urban form indicators through machine learning method

As the urban areas overcrowded, volume of urban road traffic is increasing. It could affect adverse effect on the urban resident's life quality and health. However, road traffic noise is being managed after it has affected urban residents, because it could not managed at urban planning stage. Therefore, reactive-management of road traffic noise, such as noise barrier is chosen as noise control plan, but it is costly than proactive-management. When road traffic noise could be considered as urban planning factor, the cost will be reduced. Machine learning method, such as regression analysis, decision tree and artificial neural network, could be used to analyze the relationship between variables. This method was used to analyze relationship between road traffic noise and urban form indicators. Predicted noise level of road traffic noise at building facade was used as output variable, and urban form indicators are used as input variable. Applicable urban form indicators could be divided into population-related, building-related, traffic-related and land-use-related. Building-related indicators include building coverage ratio, floor area ratio, and traffic-related indicators include traffic volume, speed and percentage of heavy vehicle. Explanation power of the prediction model derived through machine learning method could verified through correlation, coefficient of determination and k-fold cross validation.