distributions

distributions3, inspired by the eponynmous Julia package, provides a generic function interface to probability distributions. distributions has two goals:

1. Replace the rnorm(), pnorm(), etc, family of functions with S3 methods for distribution objects

2. Be extremely well documented and friendly for students in intro stat classes.

The main generics are:

• random(): Draw samples from a distribution.
• pdf(): Evaluate the probability density (or mass) at a point.
• cdf(): Evaluate the cumulative probability up to a point.
• quantile(): Determine the quantile for a given probability. Inverse of cdf().

Installation

distributions is not yet on CRAN. You can install the development version with:

install.packages("devtools")
devtools::install_github("alexpghayes/distributions3")

Basic Usage

The basic usage of distributions3 looks like:

library(distributions3)

X <- Bernoulli(0.1)

random(X, 10)
#>  [1] 0 0 0 0 0 0 0 0 0 1
pdf(X, 1)
#> [1] 0.1

cdf(X, 0)
#> [1] 0.9
quantile(X, 0.5)
#> [1] 0

Note that quantile() always returns lower tail probabilities. If you aren’t sure what this means, please read the last several paragraphs of vignette("one-sample-z-confidence-interval") and have a gander at the plot.

Contributing

I am very happy to review PRs and provide advice on how to add new functionality to the package. Documentation improvements are particularly appreciated!

To add a new distribution, the best way to get started is to look at R/Beta.R and tests/testthat/test-Beta.R, copy them, and modify them for whatever new distribution you’d like to add.

Please note that distributions3 is released with a Contributor Code of Conduct. By contributing to this project, you agree to abide by its terms.

Related work

For a comprehensive overview of the many packages providing various distribution related functionality see the CRAN Task View.

• distr is quite similar to distributions, but uses S4 objects and is less focused on documentation.
• distr6 builds on distr, but uses R6 objects
• fitdistrplus provides extensive functionality for fitting various distributions but does not treat distributions themselves as objects