multiplescca: Sparse canonical correlation vectors for fixed tuning...

In lasandrall/SELPCCA: Sparse Canonical Correlation Analysis for Associating Mutiple High Dimensional Data

Description

Obtain sparse canonical correlation vectors for fixed tuning parameters. It is recommended to use cvselpscca to choose optimal tuning paramters for each dataset, or use cvtunerange for range of tuning parameters.

Usage

multiplescca(Xdata1=Xdata1,Xdata2=Xdata2,ncancorr=ncancorr,Tau=Tau,
            CovStructure="Iden",standardize=TRUE,maxiteration=20, thresh=0.0001)

Arguments

Xdata1	A matrix of size n \times p for first dataset. Rows are samples and columns are variables.
Xdata2	A matrix of size n \times q for second dataset. Rows are samples and columns are variables.
ncancorr	Number of canonical correlation vectors. Default is one.
Tau	A vector or matrix of fixed tuning parameters for each dataset.
CovStructure	Covariance structure to use in estimating sparse canonical correlation vectors. Either "Iden" or "Ridge". Iden assumes the covariance matrix for each dataset is identity. Ridge uses the sample covariance for each dataset. See reference article for more details.
standardize	TRUE or FALSE. If TRUE, data will be normalized to have mean zero and variance one for each variable. Default is TRUE.
maxiteration	Maximum iteration for the algorithm if not converged. Default is 20.
thresh	Threshold for convergence. Default is 0.0001.

Details

The function will return three R objects, which can be assigned to a variable. To see the results, use the “$" operator.

Value

hatalpha	Estimated sparse canonical correlation vectors for first dataset.
hatbeta	Estimated sparse canonical correlation vectors for second dataset.
maxcorr	Estimated correlation from canonical correlation vectors.

References

Sandra E. Safo, Jeongyoun Ahn, Yongho Jeon, and Sungkyu Jung (2018) , Sparse Generalized Eigenvalue Problem with Application to Canonical Correlation Analysis for Integrative Analysis of Methylation and Gene Expression Data. Biometrics

See Also

cvselpscca,cvtunerange

Examples

library(SELPCCA)
##---- read in data
data(DataExample)

Xdata1=DataExample[[1]]
Xdata2=DataExample[[2]]

##---- estimate first canonical correlation vectors
ncancorr=1

#use cvtunerange for range of tuning parameters
mytunerange=cvtunerange(Xdata1=Xdata1,Xdata2=Xdata2,ncancorr=ncancorr,
                        CovStructure="Iden",standardize=TRUE)
print(mytunerange)

#Fix Tau for first and second datasets as 1.1 and 1.0 respectively
Tau=matrix(c(1,1.2,1),nrow=1)
mysparsevectors=multiplescca(Xdata1=Xdata1,Xdata2=Xdata2,ncancorr=ncancorr,
                            Tau=Tau, CovStructure="Iden",standardize=TRUE,
                            maxiteration=20, thresh=0.0001)


#example with two canonical correlation vectors
#use cvselpscca to obtain optimal tuning parameters
mycv=cvselpscca(Xdata1=Xdata1,Xdata2=Xdata2,ncancorr=ncancorr,
                CovStructure="Iden",isParallel=FALSE,ncores=NULL,nfolds=5,
                ngrid=10, standardize=TRUE,thresh=0.0001,maxiteration=20)


Tau=mycv$optTau
mysparsevectors=multiplescca(Xdata1=Xdata1,Xdata2=Xdata2,ncancorr=ncancorr,
                Tau=Tau, CovStructure="Iden",standardize=TRUE,maxiteration=20,
                thresh=0.0001)

lasandrall/SELPCCA documentation built on June 8, 2020, 12:38 a.m.