Mentor Match

This is a package to help match mentees and mentors.

It uses this implementation of Munkres to find the most effective pairings. The Munkres algorithm works on a grid of scores.

Scoring

Full details of how the matches are calculated can be read in the code itself. Customisable configurations are on the roadmap but are not planned for any upcoming releases.

Installation

You can install this project with python -m pip install mentor-match

Use

To use this library, first install it (see above). You may need to munge your data for the system to be happy with it. Use the example CSV file as guides for your mentor and mentee data, then put them together in the same folder.

The software will run as many matching exercises as you pass list[AbstractRule]. So you'll need to come up with some rules as to how you want your mentors and mentees to be matched. For more information on rules, see Rules.

Here is a snippet that outlines a minimal use in a Python project:

from matching import process
from pathlib import Path
from matching.rules.rule import Generic

data_folder = Path("Documents/mentoring-data")
mentors, mentees = process.conduct_matching_from_file(
    path_to_data=data_folder,
    rules=[[Generic({True: 3, False: 0}, lambda match: match.mentee.organisation != match.mentor.organisation)]]
)

output_folder = data_folder / "output"
process.create_mailing_list(mentors, output_folder)
process.create_mailing_list(mentees, output_folder)

This weights matches where mentors and mentees are in different organisations. For more on rules, see Rules. The system then creates a mailing list according to a set template, ready for processing by your favourite/enterprise mandated email solution

Rules

All rules are subclassed from the AbstractRule class. They need an evaluate method, which should take a Match object and return a boolean, and an apply method, which takes a Match object, evaluates it, and changes the internal state of the Match object.

I've included a couple of pre-defined rules to help start you off:

Grade

Grade needs a target difference in grades between mentors and mentees, an operator to compare them, and a score to give if the operation is true or if it's false. So Grade(2, operator.gt, {True: 3, False: 0}) will create a rule that gives 3 points to a grade difference between the mentor and the mentee that's greater than 2.

UnmatchedBonus

UnmatchedBonus only needs an integer value to add to the Match score. It'll add it if either the mentor or the mentee doesn't have any connections. This is helpful if you run multiple rounds, as it'll give the edge to mentors/mentees who haven't been successfully matched yet.

Disqualify

Disqualify needs to be passed a function that takes a Match object and returns a bool. It's an anti-rule: if the condition evaluates to True, then that Match is disqualified for this round. Two Disqualify rules are pre-defined on the Match object - a Match is disqualified if both Mentor and Mentee are the same person, or if they've already been matched once.

Generic

Generic, like Disqualify, takes a function with the signature [[Match], bool]. It also takes a dictionary, like Grade, where you define what score to be given to the Match if the function evaluates to true, or indeed if it evaluates to false!

Export

There's an inbuilt ExportToSpreadsheet which very much does what it says on the tin. Instantiate it with a list of Person objects and where you want the output to end up and call export to do the thing.

You can also subclass ExportToEmail for all your email-exporting needs. Or write your own thing entirely!