Multiple correspondence analysis - Understanding Biplots - page 394

8.9.4 Function CATPCAbipl.predregions

CATPCAbipl.predregions is used for constructing prediction regions for the
categorical variables used in a categorical PCA. It shares the following arguments with
CATPCAbipl :

Xcat orthog.transx select.origin

factor.type orthog.transy w.factor

ord.col parplotmar predict.sample

nom.col exp.factor

epsilon reverse

It has the additional argument prediction.regions (with default "all" )forspec-
ifying the categorical variable(s) for which prediction regions are to be constructed.

8.9.5 Function PCAbipl.cat

This function is to be used only with CATPCAbipl where it calls drawbipl.catPCA .

8.10 Revisiting the remuneration data: examples of MCA
and categorical PCA biplots

The remuneration data were introduced in Section 4.9.1 where all the variables used in
constructing the CVA biplots were treated as continuous except the grouping variable
Gender . The categorical variable Faclty was excluded from the analysis. As an example
of MCA we use here the 2002 data with all variables treated as nominal (unordered
categorical) while as an illustration of categorical PCA some of the variables are treated
as ordered categorical. The variables are categorized as follows:

Remun in decile categories coded as R1 , R2 , ... , R10 (highest)

Resrch categories Res0, Res1 , ... , Res7 (highest)

Rank categories Rnk1, Rnk2 , ... , Rnk5 (highest)

Age categories A1 , A2 , ... , A5 (oldest)

Gender male, female

AQual categories A1 , A2 , ... , A5 (highest)

Faclty categories F1 , F2 , ... , F9 (unordered - nominal scale)

Figures 8.16 and 8.17 contain MCA biplots based on the indicator matrix of the
dataframe Remuneration.cat.data.2002 , while the biplot in Figure 8.18 results from