Copyright | (c) 2014 Danny Navarro |
---|---|

License | BSD3 |

Maintainer | bos@serpentine.com |

Stability | experimental |

Portability | portable |

Safe Haskell | None |

Language | Haskell98 |

## Synopsis

- kruskalWallisTest :: (Ord a, Unbox a) => [Vector a] -> Maybe (Test ())
- kruskalWallisRank :: (Unbox a, Ord a) => [Vector a] -> [Vector Double]
- kruskalWallis :: (Unbox a, Ord a) => [Vector a] -> Double
- module Statistics.Test.Types

# Kruskal-Wallis test

kruskalWallisTest :: (Ord a, Unbox a) => [Vector a] -> Maybe (Test ()) #

Perform Kruskal-Wallis Test for the given samples and required
significance. For additional information check `kruskalWallis`

. This is just
a helper function.

It uses *Chi-Squared* distribution for approximation as long as the sizes are
larger than 5. Otherwise the test returns `Nothing`

.

## Building blocks

kruskalWallisRank :: (Unbox a, Ord a) => [Vector a] -> [Vector Double] #

Kruskal-Wallis ranking.

All values are replaced by the absolute rank in the combined samples.

The samples and values need not to be ordered but the values in the result are ordered. Assigned ranks (ties are given their average rank).

kruskalWallis :: (Unbox a, Ord a) => [Vector a] -> Double #

The Kruskal-Wallis Test.

In textbooks the output value is usually represented by `K`

or `H`

. This
function already does the ranking.

module Statistics.Test.Types