NA's returned by recommendHPR about rrecsys

[ludovikcoba]

If minimum = 0, you are telling the package to consider 0 as a rating, and your rating scale varies from 0 to 5. As a results the package will understand that all items are rated by all users, and will not be able to recommend any item that the user has never interacted before. If you set minimum = 1, you should be set!

[ludovikcoba]

Sure.
Try the following:

myratings <- matrix(sample(c(NA, 0:5), size = 200, replace = TRUE,  prob = c(0.6, .6,.08,.08,.08,.08,.08)), nrow = 20, byrow = TRUE)
myratings <- defineData(myratings, minimum = 0)
r <- rrecsys(myratings, alg = "FunkSVD", k = 2)
rec <- recommendHPR(r, topN=3)

NA might appear only where there are less than topN =3 unrated items.

[leo-filgueira]

So if I have a matrix where NA's are unrated itens, there is no problem setting minimum = 0 (since minimum rating is 0)?

[leo-filgueira]

Ok, I'll try.
Thanks!

[ludovikcoba]

👍