I want to run some paired comparisons on (let's say) 500 fruits. My list of fruits looks like this:
id fruit
1 Apple
2 Banana
3 Peach
...
From this list, I created all relevant pairs (without duplicates) using itertools.combinations in Python. These are >100k pairs. I added two columns for keeping track of responses:
pair fruit1 fruit2 seen decision
1 Apple Peach 0 0
2 Apple Banana 1 2
...
As you can see, the first pair has not been seen yet by the user. The second pair has been presented to the user (seen=1) and the user preferred Banana (decision=2).
This file is, obviously, rather large and a mess to work with. I was wondering whether there would be a more elegant way to do this. Ideally, I would only have the list of fruits that I showed you at the top and I would then generate the pairs on the fly during data collection. However, how do I keep track of responses and of already presented pairs without referring to the data format I am currently using?
Edit: There are some rules to the pairs.
- No "Apple vs. Apple": There cannot be comparisons between the same fruit.
- Either "Apple vs. Banana" or "Banana vs. Apple" but not both - No duplicates are allowed, even with different order.