• Stars
    star
    95
  • Rank 352,816 (Top 7 %)
  • Language
    Scala
  • License
    MIT License
  • Created almost 8 years ago
  • Updated about 2 years ago

Reviews

There are no reviews yet. Be the first to send feedback to the community and the maintainers!

Repository Details

Statistics utilities for the JVM - in Scala!

Tyche build scala version

Tyche is a small and robust statistical library for the JVM. Many JVM-hosted numerical libraries offer similar functionalities, but they seem to be unnecessarily bloated and they fail to provide straightforward APIs. Instead, Tyche is built from the ground up in accordance to good design principles.

  • Discrete & Continuous distributions.
  • Foolproof analysis tools.
  • Markov chains and lattice walks.
  • Simple Random Sampling (SRS) support.
  • Extensive documentation and good test coverage.

The library is written in Scala 2.12.0 and was tested last with sbt 0.13.13.

Behold, the power of Tyche:

// In a country in which people only want boys every family continues to
// have children until they have a boy. If they have a girl, they have
// another child. If they have a boy, they stop. What is the average
// amount of kids per family?

sealed trait Child
object Boy extends Child
object Girl extends Child

val family = tyche.DiscreteDistribution.uniform(Boy, Girl)
  .until(_ contains Boy)
  .map(_.size)

println(family.mean) // ~ 2.0

Setup

Tyche is published to Maven Central, so you just need to paste this line in you build configuration file:

libraryDependencies += "com.github.neysofu" %% "tyche" % "0.4.3"

How to contribute

Feedback and suggestions are very welcome! I invite you to check for open issues or open a fresh one to discuss around a bug or a feature idea. Please drop me a line at [email protected] if you wish you contact me.

Bonus points if you submit code!

  1. Branch off from master and start making your changes.
  2. Write a test which shows that the code works as expected.
  3. Please, please write some documentation. One or two lines per entity are enough.
  4. Send a pull request.

Alternatives

Be sure to check out some great alternatives: