predict.sgdgmf: Predict method for GMF models

In sgdGMF: Estimation of Generalized Matrix Factorization Models via Stochastic Gradient Descent

Predict method for GMF models

Description

Computes the predictions of a GMF model. Out-of-sample predictions for a new set of responses and covariates are computed via MLE, by keeping fixed the values of the estimated B and V and maximizing the likelihood with respect to A and U.

Usage

## S3 method for class 'sgdgmf'
predict(
  object,
  ...,
  newY = NULL,
  newX = NULL,
  type = c("link", "response", "terms", "coef"),
  parallel = FALSE,
  nthreads = 1
)

Arguments

object	an object of class sgdgmf
...	further arguments passed to or from other methods
newY	optionally, a matrix of new response variable
newX	optionally, a matrix of new covariate values
type	the type of prediction which should be returned
parallel	if TRUE, allows for parallel computing using the package foreach
nthreads	number of cores to be used in parallel (only if parallel=TRUE)

Details

If newY and newX are omitted, the predictions are based on the data used for the fit. In that case, the predictions corresponds to the fitted values. If newY and newX are provided, a corresponding set of A and U are estimated via maximum likelihood using the glm.fit function. By doing so, B and V are kept fixed.

Value

If type="link" or typr="response", a matrix of predictions. If type="terms", a list of matrices containing the fields XB, AZ and UV. If type="coef", a list of matrices containing the field B, A, U and V.

Examples

# Load the sgdGMF package
library(sgdGMF)

# Generate data from a Poisson model
data = sim.gmf.data(n = 120, m = 20, ncomp = 5, family = poisson())
train = sample(1:120, size = 100)
test = setdiff(1:120, train)

Y = data$Y[train, ]
newY = data$Y[test, ]

# Fit a GMF model with 3 latent factors
gmf = sgdgmf.fit(Y, ncomp = 3, family = poisson())

# Get the fitted values of a GMF model
str(predict(gmf)) # returns the overall fitted values in link scale
str(predict(gmf, type = "response")) # returns the overall fitted values in response scale
str(predict(gmf, type = "terms")) # returns the partial fitted values in link scale
str(predict(gmf, newY = newY)) # returns the predictions for the new set of responses

sgdGMF documentation built on April 3, 2025, 7:37 p.m.