Skip to content

alphanorm package

Read me

This package is related to the work in “Sparse Regularization in Marketing and Economics” by Guanhao Feng, Nicholas Polson, Yuexi Wang and Jianeng Xu. https://papers.ssrn.com/sol3/papers.cfm?abstract_id=3022856

Functions

The package contains five functions and their usage is similar to those in glmnet package.

  • alphanorm: build the alpha-norm model with input data. The coefficients are solved via a proximal algorithm and coordinate descent.
  • coef.alphanorm: take an alphanorm object as input and output the coefficients in the object
  • cv.alphanorm: use cross-validation to find the best tuning parameter $\alpha$ and $\lambda$
  • plot.alphanorm: plot the coefficient profile with respect to either $log(\lambda)$, or $l_\alpha$ norm of the coefficients
  • predict.alphanorm: get predicted value from a new input and a fitted alphanorm model

Installation

The package is currently not available on R CRAN. You can download it via:

devtools::install_github(yxwang99/alphanorm)
library(alphanorm)

Available at: https://github.com/yxwang99/alphanorm