tests/benchmarks/README.md

In quanteda/quanteda: Quantitative Analysis of Textual Data

Benchmark Guidelines

Purpose of the benchmarks/ directory

This is a where benchmarks and comparisons should exist, for comparing the performance of different methods of doing the same thing to one another in terms of performance.

It is also a good place to put comparisons of quanteda functions to similar functions from other packages, either to compare performance or behaviour.

Because the benchmarks/ directory (and its sub-directories) are specified in .Rbuildignore, none of them will be used when the package is built. So: Nothing in this directory will affect the package, and it can be used for tests and comparisons only.

Format

All benchmark files should be be in RMarkdown (.Rmd) format and compiled to .html (using knitr) prior to being pushed.

Guidelines

Naming your file

Please use one of the two filename conventions:

• benchmark_SOMETHING.Rmd - for benchmarks to compare one thing versus another
• comparison_SOMETHING.Rmd - for comparing behaviours, e.g. for comparing the behaviour of quanteda's tokenize() function to those in the tokenizers package.

Recommended benchmarking tools

• microbenchmark::microbenchmark()
• rbenchmark::benchmark()

Recommended that you use units = "relative" for more direct comparison than the actual times.

Where to put test data

In the benchmarks/data/ folder.

quanteda/quanteda documentation built on April 15, 2024, 7:59 a.m.