These are some algorithm about recommend movie

Now there are three algorithm:

• ICF predict ratings
• ICF topN recommend

• LFM predict ratings
• LFM topN recommend

We use data set of MovieLens which contain 100000 ratings from 3000+ users on 650 movies

We can see that most of the ratings are greater than 2. And popular movies have higher average rating.

First method is ICF. We try different metrics to compute the nearst neighbours.We use Pearson,jaccard,IIF and also consider time. And finally we recommend 6 movies to users. Then computer how much movie apppear in test set and the ratings are greater than 2. These are the outcome.

Use LFM to computer recommend list. We do research on the relation between #classcount, ratio and precision,recall,Fscore,Coverage.

Firstly compute genre correlation

Then use these formular to compute the recommend point of movie:

Finally,try to recommend movie to user. We compute that the precision is about 0.07, which is quite low. But we can see the concrete recommend list. We recommend movie to a kid, we can see the movie list is quite resonable.