The repo currently works by pulling in popular rating sites such as IMDB, Metacritic and Rotten Tomatoes for scores and ranking. The current algorithm currently only takes those rank orders and mixes them for a minimum score and position on those lists. There's currently zero math involved.
From a mathematical standpoint, it has proven difficult to filter outliers. A good example of this is Rambo: Last Blood has 23,428 votes on IMDB (higher than Hustlers), yet it has a rating of 26 on Metacritic. A point score system was considered but never proved fruitful.
For whoever has time, I'm open to suggestions on improving the formula. Please use this file
dump.csv.zip for experimentation.
The repo currently works by pulling in popular rating sites such as IMDB, Metacritic and Rotten Tomatoes for scores and ranking. The current algorithm currently only takes those rank orders and mixes them for a minimum score and position on those lists. There's currently zero math involved.
From a mathematical standpoint, it has proven difficult to filter outliers. A good example of this is Rambo: Last Blood has 23,428 votes on IMDB (higher than Hustlers), yet it has a rating of 26 on Metacritic. A point score system was considered but never proved fruitful.
For whoever has time, I'm open to suggestions on improving the formula. Please use this file dump.csv.zip for experimentation.