ABSTRACT
The need to introduce simultaneously quantitative and qualitative variables as active elements of one factorial analysis is common. The usual methodology is to transform the quantitative variables into qualitative variables, breaking down their variation interval into classes, and submitting the resulting homogeneous table to a multiple correspondence analysis. In 1990, Gilbert Saporta suggested introducing qualitative variables in principal component analysis (PCA) thanks to a specific metric. In reality, these two different approaches yield the same results. The resulting factor analysis presents a sufficient number of positive properties and application potential to justify the status of a separate method: factorial analysis of mixed data (FAMD). The simplest solution is to use an FAMD program such as the FAMD function of the FactoMineR package or the AFMD function in UNIWIN Plus. Otherwise, an FAMD can be performed from a regular PCA program.
