sequence (version 2.0)

ca: Correspondence analysis

Description

Achieves a Correspondence Analysis (CA) on a numeric table of class data.frame

Usage

ca(x, nfac = 3, isup = 0, jsup = 0, histev = FALSE, grr = FALSE, grc = FALSE, 
grrc = FALSE, grlist = rbind(c(1, 2), c(1, 3), c(2, 3)), prtm = FALSE, 
prtevr = FALSE, prtevc = FALSE, eps = 1e-09)

Arguments

x

data.frame minimal dimension 4 x 3. The first column must contain the character strings of the identifiers of raws any other type, class or dimension results in an error and in the program break.

nfac

Number of factors to retain (maximum 7)

isup

list of illustrative rows. 0 = no illustrative rows (default)

jsup

List of illustrative columns. Same as isup.

histev

Boolean : whether to plot or not the histogram of eigenvalues.

grr

Boolean : plot the graph of rows on the axes defined by grlist.

grc

Boolean : Plot the graph of columns on the axes defined by grlist.

grrc

Boolean : Plot the simultaneous graph of rows and columns on the axes defined by grlist. Labels of rows in black, labels of columns in red.

grlist

matrix: defines the factorial plans to plot. See details for an example.

prtm

Boolean: Print or not the data frame. Default = FALSE

prtevr

Boolean: Print or not the rows eigenvectors. Default = FALSE

prtevc

Boolean: Print or not the columns eigenvectors. Default = FALSE

eps

numeric: (tolerance) Precision for null eigenvalues. Default = 10E-09

Value

An object of class ca with attributes

fr

data.frame: weight and factorial coordinates of each row (principal and illustrative). The attribute type has the value "pri" for principal and "ill" for illustrative

fc

data.frame: weight and factorial coordinates of each column (principal and illustrative). type as in fr

Details

grlist: the successive plots to draw are defined by a matrix of dimension k,2. k = number of plans to plot. Example: to plot the plans 1-2, 1-3 and 2-3 enter sometning as matrix(1,2,1,3,2,3,nrow=3,ncol=2,byrow=2) or rbind(c(1,2),c(1,3),c(2,3)). Markovian matrix: In the case of a Markovian or of a transition matrix, one can symetrise (X + t(X)) and load it (sum of the margins added to the diagonal, before applying CA (cf See Also). In the case of a markovian square matrix (succession or transition matrix) one can symmetrize and load it (symet) before representing it by a graph (flux)

References

Van der Heijden, P. G. M. 1986. Transition matrices, model fitting and correspondence analysis. In: Data Analysis and Informatics IV (Ed. by E. Diday), pp. 221-226. Elsevier Science Publishers.

See Also

princomp, compseq to build a transition matrix, symet to modify it (symmetrization and diagonal loading), flux for the design of a graph.

Examples

# NOT RUN {
# On Csa data (xcsa)
library(sequence)
   data(xcsa)
   ca(xcsa)
# }