Evaluating boosted decision trees for billions of users

Facebook uses machine learning and ranking models to deliver the best experiences across many different parts of the app, such as which notifications to send, which stories you see in News Feed, or which recommendations you get for Pages you might want to follow. To surface the most relevant content, it's important to have high-quality machine learning models. We look at a number of real-time signals to determine optimal ranking; for example, in the notifications filtering use case, we look at whether someone has already clicked on similar notifications or how many likes the story corresponding to a notification has gotten. Because we perform this every time a new notification is generated, we want to return the decision for sending notifications as quickly as possible. More complex models can help improve the precision of our predictions and show more relevant content, but the trade-off is that they require more CPU cycles and can take longer to return results.