[Mangaki challenge] Anime and manga recommendation

Mangaki.fr is an online platform where users can (1) provide ratings of anime and manga; and (2) receive personalized recommendations of what to watch next. It is being actively developed by a community of French crazy students, and the whole code of the platform is available on GitHub, so interested users can improve the machine learning algorithms of Mangaki.

In this challenge, in order to win (and show your supremacy to the world), you will have to predict the ratings of users over unwatched anime and manga. The ratings are provided by the actual users.

You are more than welcome to look for extra sources of data in order to improve your score. All programming languages are allowed. Just find the one that works best for you! Winners will be provided an opportunity to promote their solutions on University of Big Data.

Problem type
Evaluation metric
Area under the ROC curve (AUC)
Competition status
2017/07/02 00:00 (Japan Standard Time)
2017/09/15 23:59 (Japan Standard Time)
Invitation setting
Open to everyone

Download dataset and submission

You are required to sign up and log in to downalod the dataset and make a submission.

Intermediate ranking

Intermediate rank Nickname Intermediate score
1 BC 0.80149
2 ielmansouri 0.71621
3 kazuma_desu 0.50207
4 Sablier 0.49864
5 Fistine 0.49819
6 gx 0.49751
7 admin 0.49630
7 doraemon 0.49630

This leaderboard is calculated on the latest submissions.
The intermediate scores are calculated using 50% of the test dataset, and the final scores are calculated using the other 50%.
Final ranks are determined according to the final scores.

Scores over time (Intermediate score)

Your submission timeline

You have not made any submission yet.