This script will rank the contribution of team member's contribution over a task according to the members' public voting. Ranking and weights will be generated. The algorithm is based on analytic hierarchy process (AHP), see Wikipedia for the details.

The ranking process is divided into voting and ranking. generate_tables.py will generate a voting table (.csv file) according to the related member's names. When all the members filled the tables, rank.py will give the ranking according to result in the tables.

python3 generate_tables.py [-h] --names NAMES [--output_path OUTPUT_PATH]

Theoretically, each member is responsible for filling one voting table. In the table, there is a matrix with all member names as column and row names. The matrix is initialized as all-1 matrix. Suppose the element in i th row and j th column is a_ij, the value of it should be filled with respect to the following creteria.

Theoretically, value in range (0, 9] are all acceptable and meaningful. However, adding float numbers will cause troubles when computing the reciprocal for yourself. Thus, integers from 1 to 9 is recommanded.

Note that the following relationship holds.

a_ii = 0
a_ij = 1 / a_ji

The table should be saved as .csv file.

After gathering all members' submission of voting tables into a directory, the result can be fetched by running the following command.

python3 rank.py [-h] [--input_dir INPUT_DIR] [--no_check]

Here is a brief example of using this script.

$ python3 generate_tables.py --names "Donald Trump,Влади́мир Влади́мирович Пу́тин,习近平,Theresa Mary May,Emmanuel Jean-Michel Frédéric Macron"
Generation complete. See the output files at ./vote_table.csv
After filling the tables, run rank.py

$ python3 rank.py
______________________________________________________________________
   ranking              weight                  name
         1             0.41316     Влади́мир Влади́мирович Пу́тин
         2             0.23527              Donald Trump
         3             0.21423            Theresa Mary May
         4             0.08684                  习近平
         5             0.05050  Emmanuel Jean-Michel Frédéric Macron
CI: 0.07077638131664488
----------------------------------------------------------------------

• The result of the program depends on the member's table submission quality. There is no outlier detection, thus extreme value may affect the result.
• The output of CI quantifies the consistency of the input. The closer to 0, the better result it is.
• The weights is just an reference, there is space for flexibility, such as adding an initial weight for all members and normalize the weights again.