Hacker News Reader

What is the point of predictive analytics for them? Surely they don't care about what one specific customer will do, only in aggregate? Since ML models can only learn from past aggregates, wouldn't traditional statistics on past transactions make more sense?

View on HN

They can see if marketing changes your behavior.

"We predict he won't get fries until Monday. Send a push notification."

Targeted advertising. That can be banner ads, notifications, featured suggestions or anything else. Knowing what your are most likely to accept tells them what they should offer. The goal is always conversions: turning ad impressions into sales.

Reply on news.ycombinator.com