Package 'ldatuning'

Arun2010

Description

Implement scoring algorithm

Usage

Arun2010(models, dtm)

Arguments

models	An object of class "LDA
dtm	An object of class "DocumentTermMatrix" with term-frequency weighting or an object coercible to a "simple_triplet_matrix" with integer entries.

Value

A scalar LDA model score

---

CaoJuan2009

Description

Implement scoring algorithm

Usage

CaoJuan2009(models)

Arguments

models

An object of class "LDA

Value

A scalar LDA model score

---

Deveaud2014

Description

Implement scoring algorithm

Usage

Deveaud2014(models)

Arguments

models

An object of class "LDA

Value

A scalar LDA model score

---

FindTopicsNumber

Description

Calculates different metrics to estimate the most preferable number of topics for LDA model.

Usage

FindTopicsNumber(
  dtm,
  topics = seq(10, 40, by = 10),
  metrics = "Griffiths2004",
  method = "Gibbs",
  control = list(),
  mc.cores = NA,
  return_models = FALSE,
  verbose = FALSE,
  libpath = NULL
)

Arguments

dtm	An object of class "DocumentTermMatrix" with term-frequency weighting or an object coercible to a "simple_triplet_matrix" with integer entries.
topics	Vector with number of topics to compare different models.
metrics	String or vector of possible metrics: "Griffiths2004", "CaoJuan2009", "Arun2010", "Deveaud2014".
method	The method to be used for fitting; see LDA.
control	A named list of the control parameters for estimation or an object of class "LDAcontrol".
mc.cores	NA, integer or, cluster; the number of CPU cores to process models simultaneously. If an integer, create a cluster on the local machine. If a cluster, use but don't destroy it (allows multiple-node clusters). Defaults to NA, which triggers auto-detection of number of cores on the local machine.
return_models	Whether or not to return the model objects of class "LDA. Defaults to false. Setting to true requires the tibble package.
verbose	If false (default), suppress all warnings and additional information.
libpath	Path to R packages (use only if your R installation can't find 'topicmodels' package, [issue #3](https://github.com/nikita-moor/ldatuning/issues/3). For example: "C:/Program Files/R/R-2.15.2/library" (Windows), "/home/user/R/x86_64-pc-linux-gnu-library/3.2" (Linux)

Value

Data-frame with one or more metrics. numbers of topics and corresponding values of metric. Can be directly used by FindTopicsNumber_plot to draw a plot.

Examples

## Not run: 

library(topicmodels)
data("AssociatedPress", package="topicmodels")
dtm <- AssociatedPress[1:10, ]
FindTopicsNumber(dtm, topics = 2:10, metrics = "Arun2010", mc.cores = 1L)

## End(Not run)

---

FindTopicsNumber_plot

Description

Support function to analyze optimal topic number. Use output of the FindTopicsNumber function.

Usage

FindTopicsNumber_plot(values)

Arguments

values

Data-frame with first column named 'topics' and other columns are values of metrics.

Examples

## Not run: 

library(topicmodels)
data("AssociatedPress", package="topicmodels")
dtm <- AssociatedPress[1:10, ]
optimal.topics <- FindTopicsNumber(dtm, topics = 2:10,
  metrics = c("Arun2010", "CaoJuan2009", "Griffiths2004")
)
FindTopicsNumber_plot(optimal.topics)

## End(Not run)

---

Griffiths2004

Description

Implement scoring algorithm. In order to use this algorithm, the LDA model MUST be generated using the keep control parameter >0 (defaults to 50) so that the logLiks vector is retained.

Usage

Griffiths2004(models, control)

Arguments

models	An object of class "LDA
control	A named list of the control parameters for estimation or an object of class "LDAcontrol".

Value

A scalar LDA model score

---

ldatuning: Tuning of the LDA models parameters

Description

A package for identifying the number of topics in a text corpus by generating LDA models, tuning LDA model parameters, and scoring model results.

Author(s)

Maintainer: Nathan Chaney [email protected] (ORCID) [contributor]

Authors:

• Murzintcev Nikita [email protected] (ORCID)

See Also

Useful links:

• https://github.com/nikita-moor/ldatuning

• Report bugs at https://github.com/nikita-moor/ldatuning/issues

---