This project provides rankings of IMDb movies, shows, seasons and episodes according to a combination of average ratings and number of votes.
Rankings are based on a concept previously used by IMDb to rank their top 250 rated lists described here. It follows the principle that the average ratings of titles with more votes more accurately reflect their true ratings. IMDb's approach is more sophisticated now, but they do not provide this type of ranking for sub-categories like 'genres'.
A ranking formula is applied to each category of titles (i.e. movies, series, seasons, episodes) and each genre within those categories. A season's ratings are based on aggregates of the episodes within that season (weighted average for ratings and sum for votes.)
Titles for a relevant category are ranked in descending order according to their weighted ratings, calculated as:
weighted rating = (R * v + mean * min_votes) / (v + min_votes)
where:
R = average rating for the title.
v = number of votes for the title.
min_votes = number of votes required for a more reliable rating (1 standard deviation above the mean number
of votes for the relevant category).
mean = the mean rating for the relevant category
The code for producing this data is found can be found here, allowing users to fine-tune their own rankings and do further analysis.
Data for this project is from IMDb's non-commercial dataset. A sub-set of this data is being reproduced in this project for non-commercial, educational purposes.