Collaborative Filtering in JMP/JMP Pro?

Question

I've got a customer who is interested in this analytical approach. From what I can see in JMP/JMP Pro for version 12 and what I can see in version 13 coming is there isn't much to offer in this space? Or am I mistaken? Their use case is similar to when Amazon sees you bought or viewed something and then recommends other product offerings to you.

Peter_Bartell · Answer

Thanks Scott. I did suggest to the customer they take a look at JMP Pro's clustering capability alongside some of the other modeling methods that are suggested. Because to my naïve way of looking at things clustering and some of the other mentioned multivariate techniques can be used in some contexts.

scwise · Answer

Peter,

Here are a few ideas for your customer:

Looks like there are several model based ways to get at Collaborative Filtering like Clustering, etc..

Check out nice example they have on this wikipedia page.

https://en.wikipedia.org/wiki/Collaborative_filtering

Looks like we have many of the clustering modeling capability that gets mentioned.

Listed below is an interesting step-by-step approach using clustering (in this caSE SAS Proc VarClust) that think we could replicate in JMP

http://www.dmstat1.com/res/CollaborativeFiltering.html

Lastly did see that there is some R code for Collaborative Filtering that we could integrate with in JMP.

http://www.salemmarafi.com/code/collaborative-filtering-r/

Also think Development has tried out some JMP classification type techniques at "Netflix Challenge" type problems that might make for a good reference.

Hope this helps!

Scott