Hi,
We have a requirement for a recommender system which is supposed to use collaborative filtering method. There is a solution provided with recommenderLab package. But the solution is not yet efficient to be practically used in a Spotfire document. Installing recommerderlab package to TERR also does not provide a significant change in the execution performance for the script.
Does anyone ever implemented a collaborative filtering (it can be user-based or item-based) using TERR functions without additional packages?
Any help/suggestions would be appreciated.
Kind Regards,
Ozgen.
(1) Answer
It seems that recommederlab is used frequently for this. But if you don't want to use any outside packages you would need to build everything from scratch. There are a number of walk throughts that I found, like this one (which appears to not use any outside packages):
But with the for loops seen there, I cannot comment on its performance.
And here is another one (for beer specifically, a plus for Friday afternoon here):
