Allow the user to compare candidates pairwise and record the results

[tamuri]

The tool will run a "tournament" where applicants are compared in pairs, and the results of the comparisons are used to determine overall ranking.

1. Take a list of applicants
2. Conduct pairwise comparisons between applicants
3. Rank applicants based on the comparisons

Step 1: Implement pairwise comparison logic

This should be a separate module (e.g. tournament.py) that can do the comparisons on any object (e.g. any items in a list). Don't design it specifically for the shortlister model or applicants.

Build the logic:

• Accept a list of things to compare
• Display a pair to the user
  • Create a function to generate pairs to compare
• Ask users the select the better one by typing a number or letter
• Store the results
  • Add functionality to capture and store the user's input for each pair
  • Results a saved in a format that can be looked at later
    • Print to screen, save to file
    • Allow user to save and resume comparisons

Step 2: Ranking algorithm

• Research and choose a suitable ranking method
• Write a function to process the results from step 1 and calculate the ranking
• Display the ranking to the user
• Save the ranking to file

Step 3: Integrate into shortlister tool

• Allow user to send selected applicants to compete in tournament
• Capture the ranking after the process

[shiyingwucl]

A few questions after I started to outline the code:

• What format would the data come in? (considering its not specifically for the shortlister)
• Should there be any conditions when generating the pair, or just randomly?

[tamuri]

What format would the data come in? (considering its not specifically for the shortlister)

A list of items, you don't know what they are, but they are printable: e.g. [682, 619, 807, 161] or ["lorem", "ipsum", "dolor"]

Should there be any conditions when generating the pair, or just randomly?

You need a system such that you can do the ranking - what are the options? While random is okay, you could end up getting the same pair to rank.

[tamuri]

Addition: we can assume that the input list has some kind of preliminary order that we are looking to refine. That might influence the logic to generate the pairs.