This recipe covers the steps for evaluating the output from RBM-based collaborative filtering:
- Select the ratings of a user:
inputUser = as.matrix(t(trX[75,]))
names(inputUser) <- movies_df$id_order
- Remove the movies that were not rated by the user (assuming that they have yet to be seen):
inputUser <- inputUser[inputUser>0]
- Plot the top genres seen by the user:
top_rated_movies <- movies_df[as.numeric(names(inputUser)[order(inputUser,decreasing = TRUE)]),]$Title
top_rated_genres <- movies_df[as.numeric(names(inputUser)[order(inputUser,decreasing = TRUE)]),]$Genres
top_rated_genres <- as.data.frame(top_rated_genres,stringsAsFactors=F)
top_rated_genres$count <- 1
top_rated_genres <- aggregate(count~top_rated_genres,FUN=sum,data=top_rated_genres)
top_rated_genres <- top_rated_genres[with(top_rated_genres, order(-count)), ]
top_rated_genres$top_rated_genres <- factor(top_rated_genres$top_rated_genres, levels = top_rated_genres$top_rated_genres)
ggplot(top_rated_genres[top_rated_genres$count>1,],aes(x=top_rated_genres,y=count))+
geom_bar(stat="identity")+
theme_bw()+
theme(axis.text.x = element_text(angle = 90, hjust = 1))+
labs(x="Genres",y="count",)+
theme(plot.title = element_text(hjust = 0.5))
- Reconstruct the input vector to obtain the recommendation percentages for all the genres/movies:
hh0 = tf$nn$sigmoid(tf$matmul(v0, W) + hb)
vv1 = tf$nn$sigmoid(tf$matmul(hh0, tf$transpose(W)) + vb)
feed = sess$run(hh0, feed_dict=dict( v0= inputUser, W= prv_w, hb= prv_hb))
rec = sess$run(vv1, feed_dict=dict( hh0= feed, W= prv_w, vb= prv_vb))
names(rec) <- movies_df$id_order
- Plot the top-recommended genres:
top_recom_genres <- movies_df[as.numeric(names(rec)[order(rec,decreasing = TRUE)]),]$Genres
top_recom_genres <- as.data.frame(top_recom_genres,stringsAsFactors=F)
top_recom_genres$count <- 1
top_recom_genres <- aggregate(count~top_recom_genres,FUN=sum,data=top_recom_genres)
top_recom_genres <- top_recom_genres[with(top_recom_genres, order(-count)), ]
top_recom_genres$top_recom_genres <- factor(top_recom_genres$top_recom_genres, levels = top_recom_genres$top_recom_genres)
ggplot(top_recom_genres[top_recom_genres$count>20,],aes(x=top_recom_genres,y=count))+
geom_bar(stat="identity")+
theme_bw()+
theme(axis.text.x = element_text(angle = 90, hjust = 1))+
labs(x="Genres",y="count",)+
theme(plot.title = element_text(hjust = 0.5))
- Find the top 10 recommended movies:
top_recom_movies <- movies_df[as.numeric(names(rec)[order(rec,decreasing = TRUE)]),]$Title[1:10]
The following image shows the top 10 recommended movies:
