Classification Trees for Multiple Binary Responses

Abstract

Multiple binary responses arise from many applications for which an array of health related symptoms are of primary interest. These symptoms are usually correlated. I generalize the tree-based methodology for a single response to the case involving multiple binary responses. In particular, I apply the method to analyze building-related occupant complaint syndrome based on a recently collected large database. Using the same database, I compare the performance of several node-splitting criteria for trees.

Key Words:

• Binary trees
• Indoor air quality
• Marginal models
• Model selection
• Multivariate binomial
• Nonparametric regression
• Work environment