This month we turn to Generalized Additive Models, a highly flexible yet still interpretable model.
Thank you to the AWS Loft for hosting us. The sign-in process can be time consuming so please allow extra time to enter the building. We will try to move people inside as quickly as possible.
About the Talk:
Generalized Additive Models (GAMs) model relationships between variables as flexible splines. More flexible & powerful than linear models, but more interpretable than many machine-learning approaches, GAMs are often the best choice for predicting & understanding complex nonlinear phenomena.
R's mgcv is the most popular package for fitting GAMs, but many users don't know just how versatile & powerful it is. Noam will survey the features of mgcv & demonstrate how it can be used to build GAMs that deal with a great many modeling needs: modeling continuous, count, or classification outcomes, fitting spatial data or time series, variable selection, dealing with complex hierarchical structures or fitting fully bayesian models.
Noam Ross is a disease ecologist at EcoHealth Alliance, an NGO in NYC that researches the connections between human & wildlife health. Noam builds models to understand & predict disease circulation in wildlife & spillover into people. Noam is also editor for software peer review at ROpenSci, a developer collective that builds R packages to enable open research & data. He has a Ph.D. in ecology from the University of California-Davis. Follow him on twitter at @noamross.
Pizzabegins at 6:15, the talk starts at 7, then after we head to the local bar.