Miss any of our Open RFC calls?Watch the recordings here! »

ranjs

1.23.2 • Public • Published

Build Status Coverage Status npm David David Inline docs License JavaScript Style Guide

ranjs

Statistical library for generating various seeded random variates, calculating likelihood functions and testing hypotheses (and much more).

The library includes:

  1. Statistical metrics and tests: a variety of central tendency, dispersion and shape statistics as well as statistical tests.

  2. Probability distributions: more than 130 continuous and discrete distributions (and counting), each tested rigorously for statistical correctness over a variety of parameters. Every distribution comes with the following methods:

    2.1 fast and robust sampler.
    2.2 probability density/mass function.
    2.3 cumulative distribution function.
    2.4 quantile function.
    2.5 survival, hazard and cumulative hazard functions.
    2.6 likelihood and AIC/BIC methods.
    2.7 test method that uses Kolmogorov-Smirnov test for continuous or chi2 tests for discrete distributions.

    Also, every distribution can be individually seeded.

  3. A random walk Metropolis (RWM) sampler.

install

browser

Just include the minified version and add

<script type="text/javascript" src="ran.min.js"></script>

The module will be exported under ranjs.

node

npm install ranjs

usage

distributions

const ran = require('ranjs')

// Create a new generator for Skellam distribution with mu1 = 1 and mu2 = 3
const skellam = new ran.dist.Skellam(1, 3)

// Generate 10K variates
let values = skellam.sample(1e4)

// Test if samples indeed follow the specified distribution
console.log(skellam.test(values))
// => { statistics: 14.025360669436635, passed: true }

// Evaluate PMF/CDF ...
for (let k = -10; k <= 10; k++) {
    console.log(k, skellam.pdf(k), skellam.cdf(k))
}
// => -4 0.10963424740027695 0.21542206959904264
//    -3 0.1662284357019246 0.38165050508716936
//    -2 0.20277318483535026 0.5844236896611729
//    ...

// ... or higher level statistical functions
for (let k = -4; k <= 4; k++) {
    console.log(k, skellam.hazard(k), skellam.cHazard(k))
}
// => -4 0.13973659359019766 0.24260937407418487
//    -3 0.26882602325948046 0.4807014556249526
//    -2 0.487932492278074 0.8780890224913454
//    ...


// Create another distribution and check their AIC
const skellam2 = new ran.dist.Skellam(1.2, 7.5)
console.log(`Skellam(1, 3):     ${skellam.aic(values)}`)
// => Skellam(1, 3):     41937.67252974663

console.log(`Skellam(1.2, 7.5): ${skellam2.aic(values)}`)
// => Skellam(1.2, 7.5): 66508.74299363888

demo

A demo observable notebook is available here to play around with the library.

API and documentation

For the full API and documentation, see: https://synesenom.github.io/ran/

Install

npm i ranjs

DownloadsWeekly Downloads

27

Version

1.23.2

License

MIT

Unpacked Size

605 kB

Total Files

235

Last publish

Collaborators

  • avatar