statis.dual: STATIS DUAL Analysis

Description

Implementation of the STATIS DUAL method for the joint analysis of multiple tables that share the same variables. This approach allows evaluating the common structure between tables (interstructure), building a compromise (weighted average of structures), and analyzing the trajectories of variables across the tables.

Usage

statis.dual(tables, labels.tables = NULL)

Value

A list with the following elements:

labels.tables: Vector with the table labels
interstructure: K x 2 matrix with the coordinates of the tables in the interstructure space
supervariables: p x 2 matrix with the coordinates of the variables in the compromise
trajectories: List of p K x 2 matrices, one per variable, showing its trajectory across the tables
vars.names: Names of the variables (common columns)
S: Interstructure similarity matrix
R: List of R matrices for each table
Comp: Compromise matrix (weighted combination of R matrices)
eigenvalues.compromise: Eigenvalues of the compromise (inertia per axis)
eigenvectors.compromise: Eigenvectors of the compromise
beta.weights: Weights of each table in the construction of the compromise

Arguments

tables: A list of matrices or data frames with the same columns (variables). Each element represents a "table".
labels.tables: A vector of length equal to the number of tables, used to name the tables in the results. If NULL, labels like ("T1", "T2", ...) are auto-generated.

Details

The STATIS DUAL method allows:

Evaluating structural coherence across multiple tables using the interstructure
Constructing a representative compromise of the set of tables
Analyzing the behavior of the variables across the set (trajectories)

Internally, the tables are centered and normalized considering uniform observation weights. The R matrices capture the internal structure of each table. The interstructure is based on scalar products between these matrices.

Examples

Run this code

data(Tuis5_95, Tuis5_96, Tuis5_97, Tuis5_98)
labels = c("95","96","97","98")

res <- statis.dual(list(Tuis5_95, Tuis5_96, Tuis5_97, Tuis5_98), labels.tables = labels)

# How to use res
t <- ggplot2::ggtitle("Correlation (all variables)")
plot.statis.dual.circle(list(res$supervariables), labels = row.names(res$supervariables)) + t

t <- ggplot2::ggtitle("Trajectories (all variables)")
plot.statis.dual.trajectories(res$vars.names, res$trajectories, res$labels.tables) + t