Testing low rank matrix factorization performance

recommenderlab-test RSVD

require(recommenderlab) # Install this if you don't have it already
require(devtools) # Install this if you don't have this already
# Get additional recommendation algorithms
install_github("sanealytics", "recommenderlabrats")

data(MovieLense) # Get data

# Divvy it up
scheme <- evaluationScheme(MovieLense, method = "split", train = .9,
                           k = 1, given = 10, goodRating = 4) 

scheme

# register recommender
recommenderRegistry$set_entry(
  method="RSVD", dataType = "realRatingMatrix", fun=REAL_RSVD,
  description="Recommender based on Low Rank Matrix Factorization (real data).")

# Some algorithms to test against
algorithms <- list(
  "random items" = list(name="RANDOM", param=list(normalize = "Z-score")),
  "popular items" = list(name="POPULAR", param=list(normalize = "Z-score")),
  "user-based CF" = list(name="UBCF", param=list(normalize = "Z-score",
                                                 method="Cosine",
                                                 nn=50, minRating=3)),
  "Matrix Factorization" = list(name="RSVD", param=list(categories = 10, 
                                                        lambda = 10,
                                                        maxit = 100))
)

# run algorithms, predict next n movies
results <- evaluate(scheme, algorithms, n=c(1, 3, 5, 10, 15, 20))

# Draw ROC curve
plot(results, annotate = 1:4, legend="topleft")

# See precision / recall
plot(results, "prec/rec", annotate=3)