This repository contains codes necessary to replicate Iao, Zhou and Müller (2025+): “Deep Fréchet Regression”. The DFR functions in the code folder, short for Deep Fréchet Regression, are designed for modeling the relationship between multivariate predictors and metric space-valued responses.
The folder structure of this repo is as follows:
| Folder | Detail |
|---|---|
| code | R and Python scripts for the proposed approach |
| data | Data files |
| simulation | R scripts for simulations |
| Data file. | Detail |
|---|---|
| DFR.R | Deep Fréchet Regression |
| DNN.py | Deep neural network in Python |
| NN_class.py | Class for deep neural network in Python |
| lrem.R | Local Fréchet Regression for distributional data |
| lnr.R | Local Fréchet Regression for network data |
| lcm.R | Least common multiple for a vector of integers |
| bwCV.R | Bandwidth selection for local REM using cross-validation |
| kerFctn.R | Kernel function |
| le.R | Laplacian eigenmaps function |
| dm.R | Diffusion map function |
| Data file | Detail |
|---|---|
| taxi | New York taxi data application |
| mortaility | Human mortality data applcaition |
R scripts to replicate simulation results in subsection 5.2 of the main text and Section S.2 of the Supplementary Material.
| Data file | Detail |
|---|---|
| network | Simulation for network data |
| distributional | Simulation for distributional data |
To report errors, please contact siao@ucdavis.edu. Comments and suggestions are welcome.
The Full paper can be found in "Deep Fréchet regression".
@article{iao2025dfr,
title={Deep Fr\'echet Regression},
author={Iao, Su I and Zhou, Yidong and M{\"u}ller, Hans-Georg},
journal={Journal of the American Statistical Association},
volume={120},
number={551},
pages={1437--1448},
year={2025},
publisher={Taylor \& Francis}
}