imputation (version 2.0.3)

SVTImpute: SVT Imputation

Description

Imputation using Singular Value Thresholding a la Cai, Candes, Shen.

Usage

SVTImpute(x, lambda, stepsize, threshold = 0.001, max.iters = 10, verbose = F)

Arguments

x

a data frame or matrix with size n1 x n2 where each row represents a different record

lambda

the penalty on the singular values

stepsize

optional. If not provided, uses 1.2 * (n1 * n2) / (number of missing elements)

threshold

convergence threshold

max.iters

maximum number of iterations. Note that each iteration will require computing an SVD

verbose

if TRUE print status updates

References

A Singular Value Thresholding Algorithm for Matrix Completion. Cai, Candes, Shen.

Examples

x = matrix(rnorm(100),10,10)
  x.missing = x > 1
  x[x.missing] = NA
  SVTImpute(x, 3)