News for hackers. Stuff that matters.
Re:Practical advises on recommendation systems in production by dmarchand90 on Wednesday May 04, 2022 @02:14AM (#31257271)
"Running a logistic regression might be too informed a baseline and can become a long drawn process."

What does this mean? To me it implies logistic regression is already so good that its hard to see the difference between it and most recommender systems. In which case, unless you're operating at massive scale, you can just go with logistic regression

Re:Practical advises on recommendation systems in production by Irishsteve on Wednesday May 04, 2022 @03:12AM (#31257631)
Depending on the size of the company (Revenue), and where / how the recommender system sits within the customer experience; Logistic regression could be fine. An even better baseline if discovery isn't so important for a good CX is to go with most popular and or a vendor. Building these systems used to be about getting lots of different things to work together which eats up a lot of time, and capacity (Product, UX, Eng) and it may not get the appropriate return.
Re:Practical advises on recommendation systems in production by Arkid on Wednesday May 04, 2022 @02:21AM (#31257314)
Logistic regression is a really good method and takes time to get right. But it is indeed one of the most performant models in multiple companies