predict.efh: Predict exponential family harmonium reconstruction on new...

In andland/generalizedPCA: Generalized PCA

Description

Predict exponential family harmonium reconstruction on new data

Usage

## S3 method for class 'efh'
predict(object, newdata, type = c("hidden", "link", "response"),
  ...)

Arguments

object	EFH object
newdata	matrix of the same exponential family as covariates in object.
type	the type of fitting required. type = "hidden" gives matrix of hidden mean parameters of x, type = "link" gives a matrix on the natural parameter scale, and type = "response" gives a matrix on the response scale
...	Additional arguments

Examples

# construct a low rank matrices in the natural parameter space
rows = 100
cols = 10
set.seed(1)
loadings = rnorm(cols)
mat_np = outer(rnorm(rows), rnorm(cols))
mat_np_new = outer(rnorm(rows), loadings)

# generate a count matrices
mat = matrix(rpois(rows * cols, c(exp(mat_np))), rows, cols)
mat_new = matrix(rpois(rows * cols, c(exp(mat_np_new))), rows, cols)

modp = exponential_family_harmonium(mat, k = 9, family = "poisson", max_iters = 1000)

pred = predict(modp, mat_new, type = "response")

andland/generalizedPCA documentation built on May 12, 2019, 2:42 a.m.