Semantic marking

The main objective of the semantic marking is to allow automatic extraction of information concerning the individuals belonging to a cluster.
A semanting marking is specific to a cluster and can be interpreted as a reunion of categories that best describes the individuals in the cluster.

The ENmarking() function of the EnQuireR package performs a semantic marking and provides three levels of interpretation:

  • A first one very similar to the decat() function of the FactomineR package, where the categories are taken one by one
  • A second one where the characterizing categories are considered two by two
  • A third one where the characterizing categories are considered three by three

For each level, the variables the most linked to the variable of interest are selected.

In the following example, we chose to characterize a cluster variable:

data(tea)
res.enmca=ENMCA(tea[,1:18])
res.enmark=ENmarking(dataset=res.enmca$data,var.int=ncol(res.enmca$data))
#dataset: a data frame
#var.int: the variable on which to perform the semantic marking

We get the simple variables, couples of variables and triplets of variables the most linked to each category of the variable of interest (here to each cluster).
Then a catdes() is realized on these variables, coupes of variables or triplets of variables.

Significant simple variables characterizing the first clusterClick to viewCatdes() on the simple variables characterizing the first cluster

The function GRmarking() gives a graphical representation of the semantic marking

GRmarking(res.enmark,lev=1) #res.enmark: the result of a semantic marking
#lev: level of the semantic marking (1=variables, 2=pairs and 3=triplets) to plot

Graphical representation of the first level of the semantic marking for the first cluster Click to view