So I want to rank matches, but total_count and percentage have some short comings. Something that has 1/1 match is 100%, but is less valuable that something that has 20/25 matches(even though it's a lower %). And something that has 100/5,000 matches is less valuable than something that has 20/20 matches even though the total_count is higher. Is there a good formula/equation the can be used that accounts for this?
So i've tried both total_count and pct based sorting and both work reasonably well in the middle of the data, but the top and bottom are usually not helpful.
For context. I'm trying to match Spotify playlists that have the most songs in them that a user has liked. 1 out of 1 song is not helpful. Neither is 20 out of 500. I need some mix of pct and total_count, perhaps probability of this specific match happening?.