7.2 Iterative Typology and Ascending Classification (TYPOL)

This program performs a partition of a large data set into a pre-assigned number of clusters. It creates a classification variable, summarizing a large number of variables. It is a highly versatile program; its most important features are:

  1. It can handle variables at different levels of measurement–- interval scale and nominal scale. The latter are treated as quantitative variables after their full dichotomization as binary (0,1) variables. The number of dichotomies is equal to the number of modalities of the nominal (or categorized) variables. The input variables may be quantitative, qualitative or a mix of quantitative and qualitative variables.
  2. It can handle active and passive variables. The active variables are those which are used in the construction of the typology. The passive variables are those which do not participate in the construction of the typology, but their main statistics within the typology groups are computed – mean and standard deviation for quantitative variables and frequencies for categorical variables. Thus, active variables are used to construct the typology, whereas passive variables are used to illustrate the typology.
  3. The program uses three different measures of proximity – City block distance, Euclidean distance and Benzecri’s Chi-square distance, depending upon the measurement scales of the input variables. These metrics are computed as follows: