targos / ttest Goto Github PK

#ttest

Perform the Student t hypothesis test

Installation

npm install ttest

Example

var ttest = require('ttest');

// One sample t-test
ttest([0,1,1,1], {mu: 1}).valid() // true

// Two sample t-test
ttest([0,1,1,1], [1,2,2,2], {mu: -1}).valid() // true

Documentation

var ttest = require('ttest');

The ttest module supports both one and two sample t-testing, and both equal and none equal variance.

If one array of data is given its a one sample t-test, and if two data arrays are given its a two sample t-test.

ttest() supports data in the following format:

• an array of values, e.g. ttest([1, 2, 3])
• a Summary object, e.g. ttest(new Summary([1, 2, 3]))
• an object with the following properties: mean, variance, size, e.g. ttest({mean: 123, variance: 1, size: 42})

In all cases you can also pass an extra optional object, there takes the following properties:

const options = {
  // Default: 0
  // One sample case: this is the µ that the mean will be compared with.
  // Two sample case: this is the ∂ value that the mean diffrence will be compared with.
  mu: Number,

  // Default: false
  // If true don't assume variance is equal and use the Welch approximation.
  // This only applies of two samples are used.
  varEqual: Boolean,

  // Default: 0.05
  // The significance level of the test
  alpha: Number,

  // Default "not equal"
  // What should the alternative hypothesis be:
  // - One sample case: could the mean be less, greater or not equal to mu property.
  // - Two sample case: could the mean diffrence be less, greater or not equal to mu property.
  alternative: "less" || "greater" || "not equal"
};

The t-test object is finally created by calling the ttest constructor.

const stat = ttest(sample, options);
const stat = ttest(sampleA, sampleB, options);

When the ttest object is created you can get the following information.

stat.testValue()

Returns the t value also called the statistic value.

stat.pValue()

Returns the p-value.

stat.confidence()

Returns an array containing the confidence interval, where the confidence level is calculated as 1 - options.alpha. Where the lower limit has index 0 and the upper limit has index 1. If the alternative hypothesis is less or greater one of the sides will be +/- Infinity.

stat.valid()

Simply returns true if the p-value is greater or equal to the alpha value.

stat.freedom()

Returns the degrees of freedom used in the t-test.