If an individual or organization expresses a preference between two mutually distinct alternatives, this preference can be expressed as a pairwise comparison. (Wikipedia)

If you have a set of choices and want to determine the most important one, you could use the pairwise ranking method to determine what part you should focus on.

The method consists in comparing each item from a list with each of the others, one pair at a time, based on a previously specified criterium.

In each round, one of two items is chosen an the winner is recorded. Once the entire group has undergone this process, the items are (reverse) ranked by their total number of “wins”.

Example: From the following 5 options, D is the most preferred, A the least.

A | B | C | D | E | Count | Rank | |
---|---|---|---|---|---|---|---|

A | – | B | C | D | E | 0 | 5 |

B | – | – | C | D | E | 1 | 4 |

C | – | – | – | D | C | 3 | 2 |

D | – | – | – | – | D | 4 | 1 |

E | – | – | – | – | – | 2 | 3 |

```
>>> my_list = [2,4,5,6,1,2,9]
>>> [(x, y, x >= y) for i, x in enumerate(my_list) for j, y in enumerate(my_list) if i > j]
[(4, 2, True), (5, 2, True), (5, 4, True), (6, 2, True), (6, 4, True), (6, 5, True), (1, 2, False), (1, 4, False), (1, 5, False), (1, 6, False), (2, 2, True), (2, 4, False), (2, 5, False), (2, 6, False), (2, 1, True), (9, 2, True), (9, 4, True), (9, 5, True), (9, 6, True), (9, 1, True), (9, 2, True)]
```

- Keywords: Statistics
- Quelle:
- Pairwise comparison as a tool for decision-making | 1000minds
- An example for pairwise analysis | YouTube
- This Tensorflow gives an intro to ranking algorithms and includes a short comparison of point-wise, pair-wise and list-wise ranking

- Related: ranking-algorithms