Sparse Models

Sparse methods, such as lasso regression, are attractive tools for modeling large datasets where there are many possible predictors. The limited number of non-zero predictors in such models are easier to interpret than dense methods. However, sparse models can be a poor choice when the data is generated by a system that is not truly sparse. One of the goals of the modelscape consortium is to compete modeling methods to determine best practices in application of these methods to biological and ecological applications.

Victoria DeLeo
Postodoctoral Research Associate