Factor-analysis of Flu data with BP applied within design of factor scores, as discussed in Section IV-B. The MCMC inference was based on 2000 burn-in iterations and 500 collection iterations, and factor scores are depicted for one (typical) collection sample from the Gibbs sampler. Approximately thirteen factors were inferred with non-zero factor scores (shown at right), and at left is a blow-up of the factor that most separates symptomatic (blue) from asymptomatic (red) samples. The horizontal axis denotes time in hours. The data were collected in groups, at discrete times; the results at a given time are shifted slightly along the horizontal axis with respect to one another, to enhance readability.