Parallel Sets

Parallel Sets, or ParSets (Bendix 2005, Kosara 2006), is a visualization technique that was designed specifically for interpreting categorical data. When talking to the experts analyzing the customer survey data, we realized that most of the questions they were asking were not based on individual survey responses, but on classes of answers, or sets and set intersections. How many people with more than three children under five years of age buy brand-name detergent? Or, put differently, how many members of Set A are also in Set B? How many first-class passengers on the Titanic survived (i.e., how many were in category first class on the class dimension, and in the yes category on survived)? How many of them were women (i.e., how many also had the value female in the sex dimension)?