In williamcsevier/textclass: Analytical Tool for Automated Text Analysis with Interpretable Graphical Output

knitr::opts_chunk$set(echo = TRUE)

Review Date and Time

• Review Date: 12 March 2018
• Review Time: 1320

Accuracy

Each of the functions in the textclass was evaluated using the examples provided in the vignette. Overall, the vignette provides an excellent overview of the package and its functionality.

• AFICAdata dataset
  • This dataset is well documented
  • No concerns with the dataset accuracy
• term_frequency()
  • Works as depicted in the vignette
  • if n is too large, an appropriate error message is displayed to the user
  • No concerns with term_frequency() accuracy
• tidyDTM()
  • Works as depicted in vignette
  • No concerns with tidyDTM() accuracy
• plot_topics()
  • Works as depicted in vignette
  • Performs as expected with differing values of k
  • No concerns with plot_topics() accuracy
• normalize_metrics()
  • Works as depicted in vignette
  • not clear where the values variable comes from
  • Works with different didyDTM beta values
  • No concerns with normalize_metrics accuracy
• plot_optimal_topics()
  • Works as depicted in vignette
  • No concerns with plots displayed using plot_optimal_topics()
• rank_topics()
  • Works as depicted in vignette
  • Works with different different didyDTM beta values
  • No concerns with the ranks displayed using rank_topics()

Compliation

textclass installs from github using devtools::install_github in documentation without error. The package runs with minor errors, however these errors are accompanied with explanations in the R console window.

Ease of Use

The textclass vignette makes the package easy to use. The functions in the package all have documentation that explains how they are used in this package.

Final Thoughts

I was unable to find any package-crashing errors. The code ran as expected, and error messages displayed in the console were well documented. Consider wrapping the package in a shiny app or shiny gadget to make it easier for non-programmers or individuals unfamiliar with R to use in an internet browser.

Recommended Grade: 49/50

williamcsevier/textclass documentation built on May 26, 2019, 5:36 a.m.