predict: Predict positive outcomes

In jialiwang1211/GCMlasso: Bayesian Gaussian graphical model with clustering structure for ordered variables.

Description

Making predictions for positive outcomes for binary response.

Usage

predict(GCMlasso_obj, var_response, var_group, rep_sample = 100,
  seed = 1)

Arguments

GCMlasso_obj	GCMlasso object.
var_response	index of binary response variable.
var_group	index of the variable that defines clusters.
rep_sample	number of samples in the posterior predictive sampling.
seed	a random integer.

Details

Include new observations in the data set by treating the predicting variables as NA, and run the MCMC using the function GCMlasso. Code positive cases as 1 and controls as 0. The maximum latent variable corresponding to the positive cases is used as threshold to predict new observations.

Value

Mean values of posterior predictive probabilities of positive cases for observations with known outcomes and new observations.

Examples

predict_val<-predict(GCMlasso_obj,var_response=15,var_group=16)

jialiwang1211/GCMlasso documentation built on May 14, 2019, 12:55 a.m.