[SPARK-15504] Could MatrixFactorizationModel support recommend for some users only ?

• Type: Wish
• Status: Resolved
• Priority: Trivial
• Resolution: Duplicate
• Affects Version/s: 1.6.0, 1.6.1
• Component/s: MLlib
• Labels: features performance

Environment

Spark 1.6.1

• Assignee: Unassigned
• Reporter: Hai
• Created: 24/May/16 09:41
• Updated: 25/May/16 10:51
• Resolved: 25/May/16 10:51

Description

I have used the ALS algorithm training a model, and I want to recommend products for some users not all in model, so the way I can use the API of MatrixFactorizationModel is the one -> recommendProducts(user: Int, num: Int): Array[Rating] which I should recommend the product one by one in spark driver, or the one -> recommendProductsForUsers(num: Int): RDD[(Int, Array[Rating])] which could run in spark cluster but it take some unused time calculate the user that I don't want to recommend products for. So I think if there could have an API such as -> recommendProductsForUsers(users: RDD[Int], num: Int): RDD[(Int, Array[Rating])], so it best match my case.

Issue Links

• duplicates: SPARK-10802 Let ALS recommend for subset of data