Abstract:
|
In this presentation I will discuss recent work concerning function-on-scalar regression when the number of predictors is much larger than the sample size. In particular, I will present a new methodology, called FLAME for Functional Linear Adaptive Mixed Estimation, which simultaneously selects, estimates, and smooths the important predictors in the model. Our methodology is readily available as an R package that utilizes a coordinate descent algorithm for fast implementation. Asymptotic theory will be provided and we will compare to previous methods via simulations. We will conclude by analyzing a longitudinal genetic study on childhood asthma.
|