mbmgPCA: multiblock and multigroup Principal Component Analysis

Description

multiblock and multigroup PCA (mbmgPCA)

Usage

mbmgPCA(
  Data,
  Group,
  nBlock,
  Block.name = NULL,
  ncomp = NULL,
  niter = NULL,
  ScaleGroup = FALSE,
  ScaleDataA = FALSE,
  ScaleDataB = FALSE,
  norm = FALSE
)

Arguments

Data

a numeric (quantitative) matrix or data frame

Group

a vector of factors associated with group structure

nBlock

a vector of number of variables in each block

Block.name

vector of name of blocks

ncomp

number of components, if NULL number of components is equal to min(rank(Data), M-1)

niter

number of iteration, if NULL number of iteration is equal to 10

ScaleGroup

scaling variables in each group and block, by defalt is FALSE

ScaleDataA

scaling variables in each block after group preprocessing, by defalt is FALSE

ScaleDataB

scaling variables in each block befor group preprocessing, by defalt is FALSE

norm

normalize each block, by defalt is FALSE

Value

list with the following results:

K.Data

Block data

concat.Data

Concatenated data

concat.block.Data

Block concatenated data

res.iter

Result of iteration

CRIT.h

Maximization criterion for each diemnsion

CRIT

Maximization criterion

crit.group

Maximization criterion associated with each group

crit.block

Maximization criterion associated with each block

omega

Weight of each block in construction of common scores

block.common.loading

Common loadings for each block

block.group.loadings

Partial loadings for each block and group

similarity

Similarity among common and partial loadings for each block

global.scores

Global scores among blocks

block.scores

Scores for each block

block.group.scores

Scores for each block and group

block.scores

Scores for each block

global.expvar

Global explained variance

cum.exp.var.block.group

Cumulative explained variance for each block and group

References

A. Eslami, E. M. Qannari, A. Kohler and S. Bougeard, Under Review. Multivariate data analysis of multi-groups datasets. Application to sensory analysis, Chemolab, 25, 108-123.

Examples

Run this code

# NOT RUN {
data(wine)
Select=c(which(wine[,2]=="Env1"),which(wine[,2]=="Env2"),which(wine[,2]=="Reference"))
WineData = wine[Select,-c(1,2)]
Group <- as.factor(c(rep("Env1",7), rep("Env2",5), rep("Reference",7)))
nBlock <- c(5, 3, 10, 9)
BlockNames    <- c("Olfaction at rest", "Vision", "Olfaction  after shaking", "Taste")
res = mbmgPCA(Data = WineData, Group, nBlock , Block.name=BlockNames, ncomp=5) 
# }