distriprob

Motivation

The distriprob library allows the calculation of probility density (mass in the case of discrete distributions), cumulative distribution, and quantile (a.k. inverse cdf) function values in Node or in the browser. The distriprob library is written in typescript so users of the libary can take advantage of intellisense on the module exports without any need to worry about downloading d.ts files. Plain old javascript users can also use the library, but without the benefits of typescript. The asynchronous (non-Sync) functions in the library use web workers( or webworker-threads in Node) to avoid clogging up the event loop with calculations. The library is tested against the equivalent functionality in R and every attempt is made to make the library as accurate (compared to R) and fast as possible.

Instalation and Usage

The distriprob libaray can be downloaded using NPM:

npm install distriprob

Or by cloning the github repository:

git clone https://github.com/zachmart/distriprob.git
cd distriprob
npm run build

The distriprob libary is designed to be used with nodejs or in the browser. Distriprob is written in typescript and transpiled to ES6. So the package may be imported using ES6 imports:

import * as distriprob from "distriprob";

or using commonjs require's:

const distriprob = require("distriprob");

In addition, the distriprob library contains a browserify-ed bundle for use in the browser with <script> tags which will introduce the global variable distriprob:

<script src="node_modules/distriprob/bundle.js" type="text/javascript></script>

Supported Distributions

Continuous

1. Normal (distriprob.normal)
2. Student's t (distriprob.t)
3. Chi Squared (distriprob.chi2)
4. F (distriprob.F)

Discrete

1. Binomial (distriprob.binomial)
2. Poisson (distriprob.poisson)
3. Hypergeometric (distriprob.hypergeometric)

The functionality for each of these distributions is located directly on the exported distriprob object. On each of the distribution objects there are three functions: pdf(pmf for discrete distributions), cdf, and quantile for the probability density (mass), cumulative distribution, and quantile (inverse cdf) functions respectively. Each of these functions has a synchronous version (with the "Sync" suffix) and an asynchronous version which returns an ES6 promise for the desired value. Examples:

console.log(distriprob.normal.pdfSync(0, 0, 1));   // 0.3989422804014327
distriprob.possion.quantile(0.5, 1).then((result) => {
  console.log(result);                            // 1
});

API by Distribution

Continuous Distributions

Normal

Given a random variable X with a Normal probability distribution with mean mu and standard deviation sigma:

• distriprob.normal.pdf(x, mu, sigma) returns an ES6 promise for the numeric probability density of X where:
  • x: number - is the value of X for the desired density
  • mu: number - is the mean of X, defaults to 0
  • sigma: number > 0 - is the standard deviation of X, defaults to 1
• distriprob.normal.pdfSync(x, mu, sigma) returns the numeric probability density of X where:
  • x: number - is the value of X for the desired density
  • mu: number - is the mean of X, defaults to 0
  • sigma: number > 0 - is the standard deviation of X, defaults to 1
• distriprob.normal.cdf(x, mu, sigma, lowerTail) returns an ES6 promise for the numeric cumulative distribution probability that X falls in the region delimited by the argument values below, where:
  • x: number - is the value of X bounding the region of accumulation for the desired cumulative distribution
  • mu: number - is the mean of X, defaults to 0
  • sigma: number > 0 - is the standard deviation of X, defaults to 1
  • lowerTail: boolean - determines whether the calculated cumulative distribution is for all values in the lower or upper tail (those above or below the given x)
• distriprob.normal.cdfSync(x, mu, sigma) returns the numeric cumulative distribution probability that X falls in the region delimited by the argument values below, where:
  • x: number - is the value of X bounding the region of accumulation for the desired cumulative distribution
  • mu: number - is the mean of X, defaults to 0
  • sigma: number > 0 - is the standard deviation of X, defaults to 1
  • lowerTail: boolean - determines whether the calculated cumulative distribution is for all values in the lower or upper tail (those above or below the given x)

Student's t

Chi Squared

F

Discrete Distributions

Binomial

Poisson

Hypergeometric

License

MIT --- open source