predicted_probability: Predicted Probabilities In BGGM: Bayesian Gaussian Graphical Models

Description

Compute the predicted probabilities for discrete data, with the possibility of conditional predictive probabilities (i.e., at fixed values of other nodes)

Usage

 `1` ```predicted_probability(object, outcome, Y, ...) ```

Arguments

 `object` An object of class `posterior_predict` `outcome` Character string. Node for which the probabilities are computed. `Y` Matrix (or data frame) of dimensions n (observations) by p (variables). This must include the column names. `...` Compute conditional probabilities by specifying a column name in `Y` (besides the `outcome`) and a fixed value. This can include any number of nodes. See example below. Leave this blank to compute unconditional probabilities for `outcome`.

Value

A list containing a matrix with the computed probabilities (a row for each predictive sample and a column for each category).

Note

There are no checks that the conditional probability exists, i.e., suppose you wish to condition on, say, B3 = 2 and B4 = 1, yet there is no instance in which B3 is 2 AND B4 is 1. This will result in an uninformative error.

Examples

 ``` 1 2 3 4 5 6 7 8 9 10``` ```Y <- ptsd fit <- estimate(as.matrix(Y), iter = 150, type = "mixed") pred <- posterior_predict(fit, iter = 100) prob <- predicted_probability(pred, Y = Y, outcome = "B3", B4 = 0, B5 = 0) ```

BGGM documentation built on Aug. 20, 2021, 5:08 p.m.