I haven't actually connected the front and back yet, but the big idea is Symbolic Regression. The process flow is as follows: Buoys get put in lakes and collect high resolution data. We don't really have the tools yet to understand what the data is telling us outside of the basic ecological theory. One avenue for greater understanding is to uncover the differential equations that are driving the system. In comes symbolic regression(SR), a means of deriving equations directly from observational data. Much like (non)linear regression, we want to model the data. SR searches both equation form and coefficients for fitting a data set.
This is meant to be the user interface to my master's research. I am developing a deterministic algorithm for SR that outperforms the Genetic Programming method. I want to take it one step further and have humans and computers working together to find equations or in general data mine.