Simple Statistics

simple-statistics is a Javascript library I wrote to understand statistics better. It’s useful as a node module as well as in the browser with visualization frameworks like d3.

Here are some of the kinds of things you can do:

// Find the mean (average) of a set of numbers. This
// takes an array of numbers
var mean = ss.mean([1, 2, 3]);

// The variance of numbers is the sum of the squared
// differences between numbers and the mean of the list.
var variance = ss.variance([1, 2, 3]);

// Create a linear regression based on a dataset of
// two-dimensional arrays. This returns a new function
// that you can call for the value of the line at
// each X value.
var linear_regression_line = ss.linear_regression()
    .data([[0, 1], [2, 2], [3, 3]]).line();

linear_regression_line(5);

// The r-squared function can be given a dataset just
// like linear regressions, and it'll tell you roughly how
// close the linear regression comes to actually estimating
// the data.
var r_squared = ss.r_squared([[1, 2]], linear_regression_line);

As you can see, it does a bit of descriptive statistics as well as statistical inference, and there’s even some code for a bayesian classifier in there.

Libraries like science.js, R, and descriptive-statistics are written by smarter people and are probably more performant and shiny. The point of simple-statistics is that it’s simple, accessible, and aims to be as low on concept as possible.

Install simple-statistics with npm or download simple_statistics.js from GitHub to use it in the browser.