Life Style Classification is often an important as a surrogate for more complex relationships!
Instead of building prediction models only from micro data build them also from aggregate versions of the micro data
The use of ‘building blocks’ reduces noise and represents more usually patterns of behaviour
It greatly simplifies the Data Mining task because one such derived variable is representing the average effects of many others