Modeling the Air Pollution in China – Based on Gaussian Process
Abstract
The Gaussian Process (GP) is a powerful model in machine learning that is capable to represent the distribution of functions. In this report, I first go through some basic concepts and intuitions of how the Gaussian Process works, and then bring the model to the air pollution data in China. I apply the Gaussian Process to model the air pollution from the perspective of time and geography. The result shows that the pollution level experiences di erent hour trend between northern and southern cities, and northern cities may experience more sever pollution in aggregate.