softImpute: softImpute

In jdrhyne2/FastLORS: Joint Modeling for eQTL Mapping in R

Description

softImpute is a function from Mazudmer et al. (2010). It solves the problem min || X - Z ||_Omega + alpha || Z ||_Nulear and is used in parameter tuning for LORS. Note: This function is adapted from the LORS MATLAB implementation

Usage

softImpute(X, Z, Omega0, Omega1, Omega2, alpha0, maxRank)

Arguments

X	a (possibly) incomplete matrix
Z	the target matrix
Omega0	Boolean matrix of observed entries
Omega1	Boolean matrix of training entries
Omega2	Boolean matrix of validation entries
alpha0	initial tuning parameter
maxRank	maximum rank of the solution

Value

Z	Estimate of the target matrix
Err	Squared Error of the difference between X and Z on the validation set
rank_alpha	The rank of the estimates
znorm	The sum of the soft-thresholded singular values of the estimates
Alpha	The tuning parameters used

jdrhyne2/FastLORS documentation built on March 5, 2020, 6:50 a.m.