Statistical analysis of the Tanimoto coefficient self-organizing map (TCSOM) applied to health behavioral survey data

The Tanimoto coefficient self-organizing map (TCSOM) introduced by Garavaglia (1998) is applied to self-reported data on dairy milk consumption to identify milk consumption patterns throughout a person's life. The data are a cohort of US residents aged 66 and older. These results are merged with outcomes on self-reported osteoporosis and examination data on bone mineral density (BMD). For this cohort, lack of daily milk consumption as a child and irregular patterns of milk consumption appear to contribute to osteoporosis and low BMD later in life. Simple statistical analysis, including a z-test and several empirical distribution function (EDF) tests validate some of the self-organizing properties of the TCSOM. Density plots also illustrate the nonlinear grouping and self-organizing characteristics.