Hideko wants to run cluster analysis using the data she collected in Japan. Data have 15 variables, all categorical variables including gender, age group, income group, hobby, occupation, and so forth. Keiko, one of Hidekos friends suggested that before Hideko run cluster analysis, data should be transformed to metric scale via correspondence analysis.

Hideko ran correspondence analysis and found 4 dimension solution with 78% of explained variance in the original data. Now Hideko is wondering whether she can run cluster analysis with four new variables.

What do you think about Hidekos concern? What would you advise her?