lpcr_lasso: Conduct longitudinal principal component regression analysis

In seonjoo/Lpredict: Prediction with high-dimensional longitudinal predictors using longitudinal principal component analysis

Description

Conduct longitudinal principal component regression analysis

Conduct longitudinal principal component regression analysis

Usage

lpcr_lasso(Y, Xmat = NULL, lfpca = NULL, cov = NULL,
  type.measure = "auc", T = NULL, J = NULL, I = NULL, visit = NULL,
  varthresh = 0.85, timeadjust = FALSE, nfold = 10,
  lambdalist = 2^(c(-10:10)/2), penalty.factor = NULL, M = NULL,
  seednum = 1234)

lpcr_lasso(Y, Xmat = NULL, lfpca = NULL, cov = NULL,
  type.measure = "auc", T = NULL, J = NULL, I = NULL, visit = NULL,
  varthresh = 0.85, timeadjust = FALSE, nfold = 10,
  lambdalist = 2^(c(-10:10)/2), penalty.factor = NULL, M = NULL,
  seednum = 1234)

Arguments

Y	Outcome in PCR
Xmat	Longitudinal Predictor Matrix
lfpca	(default=NULL) LFPC results. If it is not specified, the following parameters have to be specified to run LPCA
cov	Covariate
type.measure	(default= "auc")
T	Time of theimage collection
J	Total number of observations
I	Total number of subjects
visit	Vector of number of visits per subjects
varthresh	(default=0.99) Threshold for variance explained for both subject-specific and subject-visit specific compoents for dimension selection
timeadjust	(default=TRUE) Scale time per subject
verbose	(default=FALSE)
Nx	Dimension of the subject-specific components
Nw	Dimension of the subject-visit specific components
projectthresh	Threshold for variance explain in the first step of SVD
Y	Outcome in PCR
Xmat	Longitudinal Predictor Matrix
cov	Covariate
lfpca	(default=NULL) LFPC results. If it is not specified, the following parameters have to be specified to run LPCA
T	Time of theimage collection
J	Total number of observations
I	Total number of subjects
visit	Vector of number of visits per subjects
verbose	(default=FALSE)
Nx	Dimension of the subject-specific components
Nw	Dimension of the subject-visit specific components
varthresh	(default=0.99) Threshold for variance explained for both subject-specific and subject-visit specific compoents for dimension selection
projectthresh	Threshold for variance explain in the first step of SVD
timeadjust	(default=TRUE) Scale time per subject

Value

xi

phix0

phix1

zeta

phiw

nonzeroindx Non-zero index including covariates

nonzeroindx2 Non-zero index excluding covariates

refit refitted regression object

beta0 Estimated parameter for intercept

beta1 Estimated parameter for slope

auc : if type.measure = "auc", cross-validated

Author(s)

Seonjoo Lee, [e-mail]

Seonjoo Lee, [e-mail]

References

TBA

TBA

Examples

re<-hd_lfpca(Ydat,Xmat,T,J,I,visit, varthresh=0.85, timeadjust=FALSE)
lpcr_lasso(Y,lfpca=re)
re<-hd_lfpca(Ydat,Xmat,T,J,I,visit, varthresh=0.85, timeadjust=FALSE)
lpcr_lasso(Y,lfpca=re)

seonjoo/Lpredict documentation built on May 29, 2019, 6:54 p.m.