Data Challenges

Data challenges package a dataset with a defined task and an evaluation metric, which makes them useful for practice and benchmarking. The list points to the competition platforms Kaggle, DrivenData (competitions for social good), AIcrowd and Bruteforce, along with NASA's Space Apps Challenge and the ICWSM data challenge. It also links individual historic contests such as the Netflix Prize, KDD Cup 2012 by Tencent, the TravisTorrent MSR 2017 mining challenge and the Yelp dataset challenge.

Competition data usually comes with rules you accept when you join, and those rules often limit use to the competition or to non-commercial research. The Yelp dataset is for personal, educational and academic purposes. The Netflix Prize data was withdrawn after privacy concerns, so treat copies found elsewhere with care. Test labels for closed contests are often never released, so you may only be able to score against a held-out split you create yourself.