Locally Sparse Estimator for Functional Linear Regression Models

A new locally sparse (i.e., zero on some subregions) estimator for coefficient functions in functional linear regression models is developed based on a novel functional regularization technique called “fSCAD”. The nice shrinkage property of fSCAD allows the proposed estimator to locate null subregions of coefficient functions without over shrinking non-zero values of coefficient functions. Additionally, a roughness penalty is incorporated to control the roughness of the locally sparse estimator. Our method is theoretically sounder and computationally simpler than existing methods. Asymptotic analysis reveals that the proposed estimator is consistent and can identify null subregions with probability tending to one. Extensive simulations confirm the theoretical analysis and show excellent numerical performance of the proposed method. Practical merit of locally sparse modelling is demonstrated by two real applications. Supplemental materials for the paper are available online.