How wenxing99/rpedr works
This is a small research implementation of random-projection ensemble dimension reduction for high-dimensional regression. Its main workflow applies repeated low-dimensional projections, fits a base regression model, selects projections by empirical performance, and aggregates them into dimension-reduction directions. The documented follow-on algorithms estimate CMS dimension or apply the main algorithm twice. The README also identifies simulation and real-data analyses. Source excerpts were unavailable, so relationships below rely on the README; implementation details beyond it are omitted.
Main entry points: RPEDR.R RPEDR · RPEDR_dim_est.R CMS dimension estimate · RPEDR_double.R Double application · simulation_example.R Simulation example
Components 10
Reduction methods
- RPEDR
RPEDR.R - Random projections
RPEDR.R - Base regression fits
RPEDR.R - Projection selection
RPEDR.R - Reduction directions
RPEDR.R - CMS dimension estimate
RPEDR_dim_est.R - Double application
RPEDR_double.R
Research analyses
- Simulation example
simulation_example.R - Real-data analyses
Real_Data
Other components
- Researcher
Connections 9
- Researcher → RPEDR: runs
README.md - RPEDR → Random projections: applies
README.md - RPEDR → Base regression fits: fits models
README.md - RPEDR → Projection selection: selects
README.md - RPEDR → Reduction directions: aggregates by SVD
README.md - CMS dimension estimate → RPEDR: uses output
README.md - Double application → RPEDR: applies twice
README.md - Researcher → Simulation example: runs example
README.md - Researcher → Real-data analyses: reproduces analyses
README.md