https://github.com/barbehenna/colereg

compound loss minimization with nonparametric regression

https://github.com/barbehenna/colereg

Science Score: 13.0%

This score indicates how likely this project is to be science-related based on various indicators:

  • CITATION.cff file
  • codemeta.json file
  • .zenodo.json file
  • DOI references
    Found 2 DOI reference(s) in README
  • Academic publication links
  • Academic email domains
  • Institutional organization owner
  • JOSS paper metadata
  • Scientific vocabulary similarity
    Low similarity (5.3%) to scientific vocabulary
Last synced: 9 months ago · JSON representation

Repository

compound loss minimization with nonparametric regression

Basic Info
  • Host: GitHub
  • Owner: barbehenna
  • Language: R
  • Default Branch: main
  • Homepage:
  • Size: 30.3 KB
Statistics
  • Stars: 0
  • Watchers: 1
  • Forks: 0
  • Open Issues: 0
  • Releases: 0
Created about 4 years ago · Last pushed about 4 years ago

https://github.com/barbehenna/coleReg/blob/main/

# ColeReg: COmpound Loss Emporium with Regression

This project is an R "sub-package" containing nonparametric regression methods for the simultaneous estimation of normal means under squared error loss. `ColeReg` was used in the development of the methods accompanying [A nonparametric regression approach to asymptotically optimal estimation of normal means (2022)](https://doi.org/10.48550/arXiv.2205.00336); however, these methods have been integrated into [sdzhao/cole](https://github.com/sdzhao/cole) and are **maintained here for reference only**. 

## Functions

- `wtf1()` implements the constrained and penalized least-squares estimator, $\hat{d}$ in the paper. To support this function we offer two different optimizers: MOSEK and CVXR, and three different methods to select the tuning parameter: high-probability upper bound, plug-in, and cross-validation (`TV1.oracle.highprob()`, `TV1.oracle.plugin()`, and `cv.wtf1()`, respectively)
- `wtf0()` implements the monotone-only least-squares estimator given bandwidth $h$, $\tilde{d}_h$ in the paper. The bandwidth rate achieving asymptotically optimal risk is used by default. A fast, accurate approximation is used for the knots by default; however, `optimal.knots()`, gives the best knots for the risk estimate.
- `f.kde()` implements Brown and Greenshtein's $f$-estimator (https://www.jstor.org/stable/30243684) with their suggested bandwidth as default.

Owner

  • Name: Alton Barbehenn
  • Login: barbehenna
  • Kind: user
  • Company: Department of Statistics, University of Illinois Urbana-Champaign

GitHub Events

Total
Last Year