With the use of Machine learning on meteorological observation gathered from multiple sources of weather stations in Los Angeles basin, we can understand underlying causal effect relationship between meteorological parameters (such as temperature, our target features) and landuse parameters (our independant features, such as vegetation coverage, building properties,...) that would be impossilbe to extract using conventional methods. The challenge in this project is that the Landuse parameters (the independant parameters) are collinear and there are not enough datapoints to find the correct contribution of each parameter to target feature. That's when the field knowledge comes in. With the help of our expertise, we can narrow down the physical processes/mechanisms that can affect each paramter, and find the dominating parameter/mechanism that is affecting temperature.
National Science Foundation (NSF)