Fitting a mixture model to three-way data with categorical and continuous variables and missing information
by
Lyn Hunt
University of Waikato, Hamilton
Coauthors: Kaye Basford (University of Queensland, Australia)

One difficulty with all classification studies is the unobserved or missing observations that occur in data sets. In this paper, we show that the mixture likelihood approach to clustering mixed categorical and continuous three-mode three-way data can be extended to situations where some of the data are missing at random in the sense of Little and Rubin (1987). We illustrate this approach by clustering the genotypes in a three-way soybean data set where various attributes were measured on genotypes grown in several environments and where there is a moderate amount of missing data.

Date received: October 18, 2001