haozhestat / FSAM

Sparse Estimation for Functional Semiparametric Additive Models

Sparse Estimation for Functional Semiparametric Additive Models

This repository consists of coding for two simulation studies and two applications in the manuscript.

1. sim1.R: the R code for the simulation study for functional semiparametric additive model with scalar predictors;

2. sim2.R: the R code for the simulation study for functional semiparametric additive model without scalar predictors;

3. tecator.txt: the tecator data. The description for the tecator data is available on http://lib.stat.cmu.edu/datasets/tecator;

4. The folder "MatlabCode" contains one matlab file (tecator_script_tune.m), which is used to implment PACE for both training data and test data. This folder also contains a folder "pace_release2.11", which is the PACE package.

5. trainRes.mat: results from matlab codes for the training data.

6. testpace.mat: results from matlab codes for the test data.

7. Tecator analysis.R: the R code for analyzing the tecator data. We also compared different estimation methods.

8. ADHD analysis.R: the R code for analyzing the ADHD data. We also compared different estimation methods.

9. Gaussian.R, internal gau.r and iterative update.R: These three files are some R functions used for our proposed method, which is called FSAM-COSSO in our manuscript.