Simulate Mixture of Experts Linear Regression — example_simulate_bmoe • bmoe

Simulates from a VISP model and splits into training and test splits.

Usage

example_simulate_bmoe(..., multiple_y = FALSE)

example_label_switching_bmoe()

simulate_bmoe(n_s, regr, wt, prec, n_loo, q_cens = NULL)

Arguments

...: Passed to simulate_bmoe.
multiple_y: logical. When TRUE, example defaults to multiple y.
n_s: integer. Total number of simulated observations.
regr: array. Regression coefficients.
wt: matrix. Weighting coefficients.
prec: numeric. Precision parameter per component.
n_loo: integer. Number of observations to be moved into a test set.
q_cens: numeric [0, 1]. Optional artificial left-censoring level.

Value

List with class bmoe_sim.

Details

Since dim(regr) = c(n_x + 1, n_y, n_k), the dimension of that input also asserts n_x, n_y and n_k.
Component membership is simulated per observation, not per response variable. Hence, dim(wt) = c(n_x + 1, n_k).
The simulated x matrix is used for both regression and weighting.
Only n_s - n_loo of the total observations will be kept in $data.

Functions

example_label_switching_bmoe(): Reproducible example of label switching example.