febama_mcmc: Inference procedure of FEBAMA framework

In lily940703/febama: Feature-based Bayesian Forecasting Model Averaging

Inference procedure of FEBAMA framework

Description

In the inference procedure, we take the MAP estimation with the standard BFGS algorithm. If variable selection is considered (model_conf$algArgs$nIter > 1) simultaneously with MAP, we utilize the Gibbs sampler to perform variable selection over all forecasting models.

Usage

febama_mcmc(data, model_conf)

Arguments

data	A list with lpd and feat (the output of function lpd_features_multi).
model_conf	Parameter settings of FEBAMA framework. Defualt model_conf_default().

Value

febama_mcmc returns a list with the entries:

beta

A list of (number of models -1) matrices of beta in every iteration.

betaIdx

A list of (number of models -1) matrices of betaIdx in every iteration.

accept_prob

A list of (number of models -1) matrices of accept probabilities in every iteration.

lily940703/febama documentation built on March 20, 2022, 1:57 a.m.