Correspondance Analysis

Correspondance analysis provides original visualizations of factor analysis as computed using FactoMiner library in R (see how to cite this work below).

Correspondance analysis parameters

The parameters are quite intuitive.

id Fields

First one has to decide on the nature of individuals that will be considered to build the Burt matrix. For a bibliometrics analysis, one could work at the level of individuals, institutions or papers for instance.

Fields

Then the active variables should be entered (as many dimension as needed).

Supplementary Fields

Finally the last option allows to also include supplementary variables : project on the MCA but that will not take part to the definition of the axes of the factor analyze.

Number of nodes

Number of top nodes to consider in each field. Nodes with smaller values will be ignored.

Interactive visualizations produced

Three interactive visualizations are released. They are all bi-dimensional and show the positioning of variable modalities along the three first dimensions.

  1. The first html file shows dimensions 1 and 2.
  2. The second html file shows dimensions 2 and 3.
  3. The third html file show dimensions 1 and 3.

Finally the fourth html file codes first and second principal component on the x and y axes and use color for the 3rd component. Other files are produced containing the detailed positions and contributions of each variable to the correspondance analysis.

As an illustration, see the result of a correspondance analysis mixing named entities extracted from the bible:

Correspondance Analysis with 3*10 Named Entities (PER + GPE + ORG) and bible books as a supplementary field

References

Lê, S., Josse, J. & Husson, F. (2008). FactoMineR: An R Package for Multivariate Analysis. Journal of Statistical Software. 25(1). pp. 1-18.