interStructure: Inter-structure analysis.

In tuxette/SirStatis: STATIS, dual STATIS and SIR-STATIS for cubic data

Description

interStructure perfoms the inter-structure analysis and returns its result (table weights, compromise similarity matrix between tables,...). The function is also called by statis. You should only use it when wanting to perform only the inter-structure analysis.

Usage

interStructure(data, method = c("sir", "classic", "dual"), y = NULL,
  H = NULL, scale = TRUE)

Arguments

data	List; The list of tables to study.
method	The method to apply ("sir", "classic" or "dual"). Default is "sir".
y	A numerical variable of interest (mandatory for "sir").
H	Integer; The number of slices to cut y (mandatory for "sir").
scale	Logical; Should the interstructure be computed with the cosine (TRUE) or with the dot product (FALSE). Default is TRUE.

Value

S3 interstructure object; a list consisting of

• objects representative objects according to the chosen method

• corr matrix used for the eigendecomposition

• values eigenvalues of the decomposition of the RV coefficient or correlation coefficient matrix

• vectors eigenvectors of the decomposition of the RV coefficients or correlation coefficient matrix

• consensus interstructure consensual matrix

• tweights computed weights for each table

• pdata a list with two elements: pdata the preprocessed data (list with length T) used to compute the scalar products (centered and eventually scaled; conditional expectations for "sir" as produced by condExpectation) and weights the weights associated with every observation.

tuxette/SirStatis documentation built on May 14, 2019, 8:03 a.m.