Please enable JavaScript.
Coggle requires JavaScript to display documents.
TAE Week 11 - Coggle Diagram
TAE Week 11
Clustering
Hierarchical
bottom up
determine measure of dissimilarity: euclidean?
determine where to cut the dendrogram
determine linkage (ward d2)
K Means
top down - pre specify number of clusters
how many clusters to use
goal: minimise intra cluster variance and maximise inter cluster variance
collaborative and content filtering
collaborative: based on user attributes
domain free
cold start problem : how do you recommend when users have no data
predict rating based on average rating of other users
find users similar to the active user, (thru similar ratings), and then use these users ratings on other items to predict what this user will like
content: based on item attributes
no cold start problem
doesnt scale well with large datasets