LDAtools

R package for fitting topic models.

Update: We changed the name of this repo from 'LDAviz' to 'LDAtools' on 5/5/2014. It will remain a repo with tools for preprocessing raw text and fitting LDA topic models in R (with C code as the back-end to run the collapsed Gibbs sampler). Two notes:

• For visualizing the output of a topic model, please check out our repo LDAvis, hosted by Carson Sievert. All future work on visualizing topic models will be done in this repo.

• For fitting topic models, there are other software packages available, including MALLET and the R packages 'topicmodels' and 'lda', that are much more popular and better-tested (for speed and accuracy) than this package. This package was developed, more or less, (1) for practice building an R package and (2) to learn about LDA, rather than to become a widely-used package for others. So thanks for checking this out, but we'd recommend MALLET or other existing R packages for fitting topic models, and LDAvis for visualizing topic models.

Older README:

Online examples:

• xkcd comics with full analysis
• Associated Press articles

Installation:

library(devtools); install_github("LDAtools", "kshirley"); library(LDAtools)

Run AP example locally:

Make sure you have the following packages installed:

install.packages(c("plyr","reshape", "proxy", "shiny"))

library(shiny); runApp(system.file('shiny', 'hover', package='LDAtools'))

More documentation to come...