Skip to main content

Table 3 Automated model development

From: Automated Bayesian model development for frequency detection in biological time series

Models

Model ratio

H0ζ/H1ζ

1.9459e-06

H1ζ/H2ζ

1.0256e-167

H2ζ/H3ζ

622.5

H3ζ/H4ζ

566.3

H4ζ/H5ζ

501.8

H5ζ/H6ζ

99.2

  1. Posterior probability ratios of models including a different a number of background functions, ζ, in the analysis of the time series in Additional file 1, Figure S1. The first two ratios favour the more complex model (ratio below 1). The ratio between models with two and three background functions, H2ζ/H3ζ, is above 1 thus favouring the simpler model. The analysis would then normally automatically stop at two background functions, but for demonstration we include more functions here. The ratio stays above 1 thereby building a chain of decisions that always prefer the simpler model.