wls-package: Weighted least squares using C++, repeated with varying...
In wls: Iteratively Re-Weighted Least Squares

As a demonstration of C++ classes exposed through the Rcpp package, this package creates a reference class to generate repeated least-squares fits. The class is created with a response variable and a fully-specified model matrix; fits are then generated with varying weight vectors. The design is that this package can be used by any form of model-fitting that eventually uses re-weighted least squares.

This (nearly trivial) package is intended to demonstrate subclassing of the C++ class with reference classes that apply weighted least squares for a variety of purposes. The C++ under-layer takes over the numerical fitting with lower overhead in time and memory.

John Chambers <jmc at r-project.org>

wls documentation built on May 2, 2019, 6:09 p.m.