Reinforcement Learning for Personalization at Spotify with Tony Jebara - #609
The TWIML AI Podcast (formerly This Week in Machine Learning & Artificial Intelligence)
Modeling User Lifetime Value and Survival on Spotify
The value of a user's state in Spotify is determined by the sum of survival probabilities scaled by profit and a discount factor. This is known as the value function or net present value. Users are modeled with beta geometric survival models, taking into account their complex probabilities of staying or unsubscribing. Our research on user survival has been published in a 2021 paper, and we have since extended it to include multi state scenarios beyond subscriptions.
00:00
Transcript
Play full episode
Remember Everything You Learn from Podcasts
Save insights instantly, chat with episodes, and build lasting knowledge - all powered by AI.
