ExpAgendaVonmon: Bayesian hierarchical expressed agenda model

In christophergandrud/ExpAgenda: Bayesian hierarchical expressed agenda estimation using Grimmer (2010)

Description

ExpAgendaVonmon implements bayesian hierarchical topic model for texts from Grimmer (2010).

Usage

  ExpAgendaVonmon(obj = NULL, term.doc = NULL,
    authors = NULL, n.cats = 10, kappa = 400,
    verbose = TRUE)

Arguments

obj	an EADTMatrix class object created by PreProcess containing the term.doc and authors matrices. Note term.doc and authors should not be set independently if obj is specified.
term.doc	matrix. Suppose that there are D total documents and w words. term.doc is a term-document matrix with dimensions: D * w. term.doc should be sorted by author. Do not set if obj is specified.
authors	matrix. If there are n total actors whose attention to issues you would like to measure. author is an n * 2 matrix. The first column specifies the first document in the term-document matrix that was authored by a given author. The second column specifies the last document authored. Do not set if obj is specified.
n.cats	numeric. Sets the number of components (topics) in the mixture model. The default is 10.
kappa	numeric. Distribution's dispersion [CHECK].
verbose	logical. Whether or not to print out each iteration.

Value

A ExpAgendaOut object. The object contains five elements: thetas, mus, rs, alpha, and authorID. thetas are [FILL IN]. mus are location on the unit hyperspace where the vMF distribution reaches its mode for each stem [CHECK]. rs are the probability of document j from author i being from topic k. alphas are the prior distributions of the topics. authorID is used for DocTopics to return the documents their their original order.

Source

Grimmer, J. (2010). A Bayesian Hierarchical Topic Model for Political Texts: Measuring Expressed Agendas in Senate Press Releases. Political Analysis, 18, 1-35. http://pan.oxfordjournals.org/content/18/1/1.short.

christophergandrud/ExpAgenda documentation built on May 13, 2019, 7:01 p.m.