Fit a Bayesian Bradley–Terry or Plackett–Luce model

fit_btsbm() is the common fitting entry point. This first implementation supports simple BT, BT–SBM, simple PL, PL–SBM, PL ranking mixtures, and PL–LBM. The latter two start with readable R Gibbs kernels derived from the PLuce exponential-race architecture; their high-performance C++ kernels are a later parity-tested optimisation.

Usage

fit_btsbm(data, model, control = mcmc_control())

Arguments

data

A btsbm_pairwise_data or btsbm_ranking_data object created by as_bt_data() or as_rankings().

model

A btsbm_model object.

control

A btsbm_mcmc_control object.

Value

An object of class btsbm_fit.

References

Caron, F. and Doucet, A. (2012). Efficient Bayesian inference for generalized Bradley–Terry models. Journal of Computational and Graphical Statistics, 21(1), 174–196. doi:10.1080/10618600.2012.638220