Figure 1.

Discretization of data for a hypothetical National Health and Nutrition Examination Survey (NHANES) subsample consisting of five subjects (S1–S5) that were each tested for six chemicals (C1–C6). The concentration data consist of real numbers representing concentrations, whereas the presence–absence data consist of binary digits, with 1 indicating presence and 0 indicating absence. For each chemical concentration, the appropriate discretization threshold was used to determine presence or absence. For example, the concentration in the top left cell of the concentration data matrix (0.63) was converted to a 1 in the presence–absence data matrix because 0.63 exceeds the chemical-specific threshold of 0.4.