This function recovers parameter estimates of the multinomial topic model given parameter estimates for a Poisson non-negative matrix factorization (NMF).

poisson2multinom(fit)

Arguments

fit

An object of class “poisson_nmf_fit”, such as an output from fit_poisson_nmf. It does not make sense for a multinomial topic model to have less than two topics, so an error will be reported when k < 2, where k is the rank of the matrix factorization. If a multinomial topic model fit is provided (that is, an object of class “multinom_topic_model_fit”), the fit object is immediately returned “as is”.

Value

The return value is the list fit, in which fit$F and fit$L are the parameters of the multinomial topic model; specifically, fit$L[i,] gives the topic probabilities for sample or document i, and fit$F[,k] gives the term probabilities for topic k. An additional vector fit$s of length n is returned giving the "size factors".