sccoda.model.scCODA_model.CompositionalModel.get_chains_after_burnin

CompositionalModel.get_chains_after_burnin(samples, kernel_results, num_burnin, is_nuts=False)

Application of burn-in after MCMC sampling. Cuts the first num_burnin samples from all inferred variables and diagnostic statistics.

Parameters

samples : List[any]List[any]

all kernel states

kernel_results : List[any]List[any]

Kernel meta-information. The tracked statistics depend on the sampling method.

num_burnin : intint

number of burn-in iterations

is_nuts : boolbool (default: False)

Specifies whether NUTS sampling was used

Return type

Tuple[List[any], dict, float]Tuple[List[any], dict, float]

Returns

• MCMC chain without burn-in, sampling statistics, acceptance rate

• states_burnin – Kernel states without burn-in samples

• stats – sampling statistics

• p_accept – acceptance rate of MCMC process