missing_data_model: Sample missing data

In deruncie/MegaLMM: MegaLMM

Sample missing data

Description

Function to sample missing data given model parameters.

Usage

missing_data_model(observation_model_parameters, MegaLMM_state = list())

Arguments

observation_model_parameters	List of parameters necessary for the data model. Here, a Matrix of coordinates of NAs in Y
MegaLMM_state	a MegaLMM_state object. Generally, only current_state and data_matrices is used. If empty, will return a default set of parameters of the appropriate size for model initialization.
Y	data matrix n_Y x p_Y

Details

This is also a template for data_model functions.

The function should draw a posterior sample for Eta (the (potentially) latent data on the linear scale) given the factor model and the observed data. It returns a list with Eta and any other variables associated with the data_model. These variables are added to current_state, but are not currently saved in Posterior

Initial values, hyperparameters, and any helper functions / parameters can be passed in observation_model_parameters.

When run with an empty current_state and data_matrices == NULL, it should return a matrix Eta of the correct dimensions, but the values are unimportant.

Value

list of data_model variables including:

state a list of parameters associated with the data_model. Here, only the matrix Eta

posteriorSample_params a list of parameter names to record posterior samples of

posteriorMean_params a list of parameters to record the posterior mean, but not save individual posterior samples

deruncie/MegaLMM documentation built on June 10, 2025, 7:26 p.m.