Predict the mean matrix from a fitted generalized principal component analysis model object.
# S3 method for glmpca
predict(object, ...)a fitted object of class inheriting from glmpca.
additional named arguments. Currently ignored.
a dense matrix of predicted mean values.
The predicted mean matrix returned by this function will have the same dimensions as the original data matrix and it will be dense even if the original data were sparse. This can exhaust available memory for large datasets, so use with caution.
Let Y be the data matrix originally used to estimate the
parameters in fit. The GLM-PCA model regards each element of
Y as a random sample from an exponential family distribution
such as a Poisson, negative binomial, or binomial likelihood. The
components of a GLM-PCA fit are combined to produce the predicted
mean of this distribution at each entry of Y. This matrix may be
regarded as a 'denoised' version of the original data.
glmpca,
predict.glm with type='response'