msma: Multiblock Sparse Partial Least Squares

In msma: Multiblock Sparse Multivariable Analysis

Multiblock Sparse Partial Least Squares

Description

This is a function for a matrix decomposition method incorporating sparse and supervised modeling for a multiblock multivariable data analysis

Usage

msma(X, ...)

## Default S3 method:
msma(
  X,
  Y = NULL,
  Z = NULL,
  comp = 2,
  lambdaX = NULL,
  lambdaY = NULL,
  lambdaXsup = NULL,
  lambdaYsup = NULL,
  eta = 1,
  type = "lasso",
  inX = NULL,
  inY = NULL,
  inXsup = NULL,
  inYsup = NULL,
  muX = 0,
  muY = 0,
  defmethod = "canonical",
  scaling = TRUE,
  verbose = FALSE,
  intseed = 1,
  ceps = 1e-04,
  ...
)

## S3 method for class 'msma'
print(x, ...)

Arguments

X	a matrix or list of matrices indicating the explanatory variable(s). This parameter is required.
...	further arguments passed to or from other methods.
Y	a matrix or list of matrices indicating objective variable(s). This is optional. If there is no input for Y, then PCA is implemented.
Z	a vector, response variable(s) for implementing the supervised version of (multiblock) PCA or PLS. This is optional. The length of Z is the number of subjects. If there is no input for Z, then unsupervised PLS/PCA is implemented.
comp	numeric scalar for the maximum number of componets to be considered.
lambdaX	numeric vector of regularized parameters for X, with a length equal to the number of blocks. If lambdaX is omitted, no regularization is conducted.
lambdaY	numeric vector of regularized parameters for Y, with a length equal to the number of blocks. If lambdaY is omitted, no regularization is conducted.
lambdaXsup	numeric vector of regularized parameters for the super weight of X with length equal to the number of blocks. If omitted, no regularization is conducted.
lambdaYsup	numeric vector of regularized parameters for the super weight of Y with length equal to the number of blocks. If omitted, no regularization is conducted.
eta	numeric scalar indicating the parameter indexing the penalty family. This version contains only choice 1.
type	a character, indicating the penalty family. In this version, only one choice is available: "lasso."
inX	a vector or list of numeric vectors specifying the variables in X, always included in the model
inY	a vector or list of numeric vectors specifying the variables in Y, always included in the model
inXsup	a (list of) numeric vector to specify the blocks of X which are always in the model.
inYsup	a (list of) numeric vector to specify the blocks of Y which are always in the model.
muX	a numeric scalar for the weight of X for the supervised case. 0 <= muX <= 1.
muY	a numeric scalar for the weight of Y for the supervised case. 0 <= muY <= 1.
defmethod	a character representing the deflation method. This version has only the choice "canonical."
scaling	a logical, indicating whether or not data scaling is performed. The default is TRUE.
verbose	information
intseed	seed number for the random number in the parameter estimation algorithm.
ceps	a numeric scalar for the convergence condition of the algorithm
x	an object of class "msma", usually, a result of a call to msma

Details

msma requires at least one input X (a matrix or list). In this case, (multiblock) PCA is conducted. If Y is also specified, then a PLS is conducted using X as explanatory variables and Y as objective variables. This function scales each data matrix to a mean of 0 and variance of 1 in the default. The block structure can be represented as a list. If Z is also specified, a supervised version is implemented, and the degree is controlled by muX or muY, where 0 <= muX <= 1, 0 <= muY <= 1, and 0 <= muX + muY < 1. If a positive lambdaX or lambdaY is specified, then a sparse estimation based on the L1 penalty is implemented.

Value

dmode	Which modes "PLS" or "PCA"
X	Scaled X which has a list form.
Y	Scaled Y which has a list form.
Xscale	Scaling information for X. The means and standard deviations for each block of X are returned.
Yscale	Scaling information for Y. The means and standard deviations for each block of Y are returned.
comp	the number of componets
wbX	block loading for X
sbX	block score for X
wbY	block loading for Y
sbY	block score for Y
ssX	super score for X
wsX	super loading for X
ssY	super score for Y
wsY	super loading for Y
nzwbX	number of nonzeros in block loading for X
nzwbY	number of nonzeros in block loading for Y
nzwsX	number of nonzeros in super loading for X
nzwsY	number of nonzeros in super loading for Y
selectXnames	names of selected variables for X
selectYnames	names of selected variables for Y
avX	the adjusted variance of the score for X
avY	the adjusted variance of the score for Y
cpevX	the cumulative percentage of the explained variance for X
cpevY	the cumulative percentage of the explained variance for Y
reproduct	Predictivity. Correlation between Y and the predicted Y
predictiv	Reproductivity. Correlation between the score for Y and the outcome Z

Examples

##### data #####
tmpdata = simdata(n = 50, rho = 0.8, Yps = c(10, 12, 15), Xps = 20, seed=1)
X = tmpdata$X; Y = tmpdata$Y 

##### One Component #####
fit1 = msma(X, Y, comp=1, lambdaX=2, lambdaY=1:3)
fit1

##### Two Component #####
fit2 = msma(X, Y, comp=2, lambdaX=2, lambdaY=1:3)
fit2

##### Sparse Principal Component Analysis #####
fit3 = msma(X, comp=5, lambdaX=2.5)
summary(fit3)

msma documentation built on May 29, 2024, 2:52 a.m.