mann-whitney-utest
Version:
The Mann-Whitney U test is a nonparametric statistical test
39 lines (28 loc) • 1.67 kB
Markdown
# Mann-Whitney U Test
[](https://travis-ci.org/lukem512/mann-whitney-utest)  [](https://www.npmjs.com/package/mann-whitney-utest) [](https://www.npmjs.com/package/mann-whitney-utest) [](https://www.npmjs.com/package/mann-whitney-utest)
This is an NPM module that allows you to perform the Mann-Whitney U test on numeric samples. The Mann-Whitney U test is a nonparametric statistical test that does not assume a normal distribution.
To use it, simply install via NPM and include it in your project file.
```
var mwu = require('mann-whitney-utest');
```
Then, to test an array of samples, use the `test` method.
```
var samples = [ [30, 14, 6], [12, 15, 16] ];
console.log(mwu.test(samples)); // [ 4, 5 ]
```
To test whether the result is significant, use the `significant` method. This tests the U-value against an approximate critical value.
```
var u = mwu.test(samples);
if (mwu.significant(u, samples)) {
console.log('The data is significant!');
} else {
console.log('The data is not significant.');
}
```
You can check your answers using the `check` method. This exploits a property of the Mann-Whitney test that ensures the sum of the U values does not exceed the product of the number of observations.
```
var u = mwu.test(samples);
if (mwu.check(u, samples)) {
console.log('The values are correct');
}
```