Cookies help us deliver our services. By using our services, you agree to our use of cookies. Learn more

crowdAI is shutting down - please read our blog post for more information

Spotify <> crowdAI : How likely are you to skip a particular song ?

16 November 2018


Image Source: Wikipedia

You open up your favorite music streaming app, and you have a very interesting playlist on shuffle.

The first song starts playing, you don’t really like it, so you skip it.
The next song plays, but you heard it 3 times in a row on your evening jog, you skip it.
The next song plays, it’s an unknown artist who hasn’t tickled your interest yet, so you skip it.

Before you end up with a song that has all your attention, you already have skipped quite a few songs. And you are not alone: about ~24% of the songs are skipped in just the first 5 seconds of the song. And about ~48% of the songs will be skipped before the end of the song! What influences this skipping behavior? The previous tracks you listened to? The contents of the actual track? The listener’s age? The listener’s gender? The day of the week? All of them?

This winter, we team up with Spotify in order to precisely answer that question.

As part of the Spotify Skip Prediction Challenge on crowdAI, we release a public dataset of ~130 million listening sessions (with associated user interactions) from Spotify’s online music streaming service. The task is to predict whether individual tracks encountered in a listening session will be skipped by a particular user.

In order to do this, complete information about the first half of a user’s listening session is provided, while the prediction is to be carried out on the second half. Participants have access to metadata, as well as acoustic descriptors, for all the tracks encountered in listening sessions.

We will evaluate your submissions on a private test set of about ~30 million listening sessions. Looking at the information you have about the first half of a listening session (and any other information you choose to aggregate about a particular user, or the current track), your task is to simply predict whether or not the track in question will be skipped by the user in the second half.

A simple 0 or a 1.

The top participants will be invited to present their results at a workshop on February 15th, held as a part of WSDM 2019, Melbourne, Australia. The top 3 participants will also be awarded cash prizes of AUD 2000, AUD 750 and AUD 250 respectively. The bragging rights for impacting the music stream experiences for millions of Spotify users come for free. :wink:

This is the first time Spotify comes to the crowdAI community with a really interesting problem, and a much more interesting dataset. Let’s show them some love!

We’re looking forward to a really great competition, and the trademarked enthusiasm from the crowdAI community!

(on behalf of the competition organizers)