realRatingMatrix {recommenderlab}

Class "realRatingMatrix": Real-valued Rating Matrix

Description

A matrix containing ratings (typically 1-5 stars, etc.).

Objects from the Class

Objects can be created by calls of the form new("realRatingMatrix", data = m), where m is sparse matrix of class dgCMatrix in package Matrix or by coercion from a regular matrix, a data.frame containing user/item/rating triplets as rows, or a sparse matrix in triplet form (dgTMatrix in package Matrix).

Slots

data:

Object of class "dgCMatrix", a sparse matrix defined in package Matrix. Note that this matrix drops NAs instead of zeroes. Operations on "dgCMatrix" potentially will delete zeroes.

normalize:

NULL or a list with normalizaton factors.

Extends

Class "ratingMatrix", directly.

Methods

coerce

signature(from = "matrix", to = "realRatingMatrix"): Note that unknown ratings have to be encoded in the matrix as NA and not as 0 (which would mean an actual rating of 0).

coerce

signature(from = "realRatingMatrix", to = "matrix")

coerce

signature(from = "data.frame", to = "realRatingMatrix"): coercion from a data.frame with three columns. Col 1 contains user ids, col 2 contains item ids and col 3 contains ratings.

coerce

signature(from = "realRatingMatrix", to = "data.frame"): produces user/item/rating triplets.

coerce

signature(from = "realRatingMatrix", to = "dgTMatrix")

coerce

signature(from = "dgTMatrix", to = "realRatingMatrix")

coerce

signature(from = "realRatingMatrix", to = "dgCMatrix")

coerce

signature(from = "dgCMatrix", to = "realRatingMatrix")

coerce

signature(from = "realRatingMatrix", to = "ngCMatrix")

binarize

signature(x = "realRatingMatrix"): create a "binaryRatingMatrix" by setting all ratings larger or equal to the argument minRating as 1 and all others to 0.

getTopNLists

signature(x = "realRatingMatrix"): create top-N lists from the ratings in x. Arguments are n (defaults to 10), randomize (default is NULL) and minRating (default is NA). Items with a rating below minRating will not be part of the top-N list. randomize can be used to get diversity in the predictions by randomly selecting items with a bias to higher rated items. The bias is introduced by choosing the items with a probability proportional to the rating (r-min(r)+1)^{randomize}. The larger the value the more likely it is to get very highly rated items and a negative value for randomize will select low-rated items.

removeKnownRatings

signature(x = "realRatingMatrix"): removes all ratings in x for which ratings are available in the realRatingMatrix (of same dimensions as x) passed as the argument known.

rowSds

signature(x = "realRatingMatrix"): calculate the standard deviation of ratings for rows (users).

colSds

signature(x = "realRatingMatrix"): calculate the standard deviation of ratings for columns (items).

See Also

See ratingMatrix inherited methods, binaryRatingMatrix, topNList, getList and getData.frame. Also see dgCMatrix, dgTMatrix and ngCMatrix in Matrix.

Examples

## create a matrix with ratings
m <- matrix(sample(c(NA,0:5),100, replace=TRUE, prob=c(.7,rep(.3/6,6))),
	nrow=10, ncol=10, dimnames = list(
	    user=paste('u', 1:10, sep=''),
	    item=paste('i', 1:10, sep='')
    ))
m

## coerce into a realRatingMAtrix
r <- as(m, "realRatingMatrix")
r

## get some information
dimnames(r)
rowCounts(r) ## number of ratings per user
colCounts(r) ## number of ratings per item
colMeans(r) ## average item rating
nratings(r) ## total number of ratings
hasRating(r) ## user-item combinations with ratings

## histogram of ratings
hist(getRatings(r), breaks="FD")

## inspect a subset
image(r[1:5,1:5])

## coerce it back to see if it worked
as(r, "matrix")

## coerce to data.frame (user/item/rating triplets)
as(r, "data.frame")

## binarize into a binaryRatingMatrix with all 4+ rating a 1
b <- binarize(r, minRating=4)
b
as(b, "matrix")