4. Experiment 3: ${\rm IVGA_{VQ}}$(A+B+C).

The feature sets A, B and C were concatenated and ${\rm IVGA_{VQ}}$ was applied for modeling and grouping. The results are presented in Table 1. Twelve groups were found, and in terms of grouping the result was excellent: each group contained variables from one variable category only. Furthermore, the improved model cost shows that presented ${\rm IVGA_{VQ}}$ algorithm performs well in the IVGA task.