| public Predict ( int user_id, int item_id ) : double | ||
| user_id | int | |
| item_id | int | |
| 리턴 | double | |
public static List<Movie> GetRecommendations(List<Movie> all, Dictionary<int, int> newRanks, List<Tuple<int, int, float>> oldRanks)
{
int newUserId = 0;
Ratings ratings = new Ratings();
foreach (var r in oldRanks)
{
ratings.Add(r.Item1, r.Item2, r.Item3);
if (r.Item1 > newUserId) newUserId = r.Item1;
}
// this makes us sure that the new user has a unique id (bigger than all other)
newUserId = newUserId + 1;
foreach (var k in newRanks)
{
ratings.Add(newUserId, k.Key, (float)k.Value);
}
var engine = new BiPolarSlopeOne();
// different algorithm:
// var engine = new UserItemBaseline();
engine.Ratings = ratings;
engine.Train(); // warning: this could take some time!
return all.Select(m =>
{
m.Rank = engine.Predict(newUserId, m.Id); // do the prediction!
return m;
}).ToList();
}
public void TestNewUserInTestSet()
{
var recommender = new BiPolarSlopeOne();
var training_data = new Ratings();
training_data.Add(0, 0, 1.0f);
training_data.Add(1, 1, 5.0f);
recommender.Ratings = training_data;
recommender.Train();
Assert.AreEqual( 3.0, recommender.Predict(2, 1) );
}
public void TestNewItemInTestSet()
{
var recommender = new BiPolarSlopeOne();
recommender.MinRating = 1;
recommender.MaxRating = 5;
var training_data = new Ratings();
training_data.Add(0, 0, 1.0);
training_data.Add(1, 1, 5.0);
recommender.Ratings = training_data;
recommender.Train();
Assert.AreEqual( 3.0, recommender.Predict(0, 2) );
}