public void testFullCorrelation1Weighted()
{
IDataModel dataModel = getDataModel(
new long[] {1, 2},
new Double?[][] {
new double?[]{3.0, -2.0},
new double?[]{3.0, -2.0},
});
double correlation = new PearsonCorrelationSimilarity(dataModel, Weighting.WEIGHTED).UserSimilarity(1, 2);
assertCorrelationEquals(1.0, correlation);
}
public void testFullCorrelation1()
{
IDataModel dataModel = getDataModel(
new long[] { 1, 2 },
new Double?[][] {
new double?[] { 3.0, -2.0 },
new double?[] { 3.0, -2.0 },
});
double correlation = new PearsonCorrelationSimilarity(dataModel).UserSimilarity(1, 2);
assertCorrelationEquals(1.0, correlation);
}
public void testNoCorrelation1Weighted()
{
IDataModel dataModel = getDataModel(
new long[] { 1, 2 },
new Double?[][] {
new double?[] { 3.0, -2.0 },
new double?[] { -3.0, 2.0 },
});
double correlation = new PearsonCorrelationSimilarity(dataModel, Weighting.WEIGHTED).UserSimilarity(1, 2);
assertCorrelationEquals(-1.0, correlation);
}
public void testNoCorrelation3()
{
IDataModel dataModel = getDataModel(
new long[] { 1, 2 },
new Double?[][] {
new double?[] { 90.0, 80.0, 70.0 },
new double?[] { 70.0, 80.0, 90.0 },
});
double correlation = new PearsonCorrelationSimilarity(dataModel).UserSimilarity(1, 2);
assertCorrelationEquals(-1.0, correlation);
}
public void testFullCorrelation2()
{
IDataModel dataModel = getDataModel(
new long[] {1, 2},
new Double?[][] {
new double?[]{3.0, 3.0},
new double?[]{3.0, 3.0},
});
double correlation = new PearsonCorrelationSimilarity(dataModel).UserSimilarity(1, 2);
// Yeah, undefined in this case
Assert.True(Double.IsNaN(correlation));
}
public void testSimple()
{
IDataModel dataModel = getDataModel(
new long[] { 1, 2 },
new Double?[][] {
new double?[] { 1.0, 2.0, 3.0 },
new double?[] { 2.0, 5.0, 6.0 },
});
double correlation = new PearsonCorrelationSimilarity(dataModel).UserSimilarity(1, 2);
assertCorrelationEquals(0.9607689228305227, correlation);
}
public void testSimpleWeighted()
{
IDataModel dataModel = getDataModel(
new long[] { 1, 2 },
new Double?[][] {
new double?[] { 1.0, 2.0, 3.0 },
new double?[] { 2.0, 5.0, 6.0 },
});
double correlation = new PearsonCorrelationSimilarity(dataModel, Weighting.WEIGHTED).UserSimilarity(1, 2);
assertCorrelationEquals(0.9901922307076306, correlation);
}
public void testNoCorrelation2()
{
IDataModel dataModel = getDataModel(
new long[] { 1, 2 },
new Double?[][] {
new double?[] { null, 1.0, null },
new double?[] { null, null, 1.0 },
});
double correlation = new PearsonCorrelationSimilarity(dataModel).UserSimilarity(1, 2);
Assert.True(Double.IsNaN(correlation));
}
public void testNoItemCorrelation1()
{
IDataModel dataModel = getDataModel(
new long[] { 1, 2 },
new Double?[][] {
new double?[] { 3.0, -3.0 },
new double?[] { 2.0, -2.0 },
});
double correlation =
new PearsonCorrelationSimilarity(dataModel).ItemSimilarity(0, 1);
assertCorrelationEquals(-1.0, correlation);
}
public void testFullCorrelation2()
{
IDataModel dataModel = getDataModel(
new long[] { 1, 2 },
new Double?[][] {
new double?[] { 3.0, 3.0 },
new double?[] { 3.0, 3.0 },
});
double correlation = new PearsonCorrelationSimilarity(dataModel).UserSimilarity(1, 2);
// Yeah, undefined in this case
Assert.True(Double.IsNaN(correlation));
}
public void testSimpleItemWeighted()
{
IDataModel dataModel = getDataModel(
new long[] { 1, 2, 3 },
new Double?[][] {
new double?[] { 1.0, 2.0 },
new double?[] { 2.0, 5.0 },
new double?[] { 3.0, 6.0 },
});
IItemSimilarity itemSimilarity = new PearsonCorrelationSimilarity(dataModel, Weighting.WEIGHTED);
double correlation = itemSimilarity.ItemSimilarity(0, 1);
assertCorrelationEquals(0.9901922307076306, correlation);
}
public Result RecommendPearsonCorrelationSimilarity(int userId, int movieId)
{
Result result = new Result();
dataModel = new FileDataModel(PathToDataFile, false, FileDataModel.DEFAULT_MIN_RELOAD_INTERVAL_MS, false, ",", userId, movieId);
var removedPrefs = GenericDataModel.preferenceFromUsersRemoved.Values;
var valueToCompare = removedPrefs.FirstOrDefault(i => i.GetItemID() == movieId).GetValue();
var similarity = new PearsonCorrelationSimilarity(dataModel);
var recommender = new GenericItemBasedRecommender(dataModel, similarity);
var preferences = recommender.EstimatePreference(userId, movieId);
result.PredictedValue = preferences;
result.RealValue = removedPrefs.First().GetValue();
return result;
}
public void testFromCorrelation()
{
IDataModel dataModel = getDataModel(
new long[] {1, 2, 3},
new Double?[][] {
new double?[] {1.0, 2.0},
new double?[] {2.0, 5.0},
new double?[] {3.0, 6.0},
});
IItemSimilarity otherSimilarity = new PearsonCorrelationSimilarity(dataModel);
IItemSimilarity itemSimilarity = new GenericItemSimilarity(otherSimilarity, dataModel);
assertCorrelationEquals(1.0, itemSimilarity.ItemSimilarity(0, 0));
assertCorrelationEquals(0.960768922830523, itemSimilarity.ItemSimilarity(0, 1));
}
public void testInferrer()
{
IDataModel dataModel = getDataModel(
new long[] { 1, 2 },
new Double?[][] {
new double?[] { null, 1.0, 2.0, null, null, 6.0 },
new double?[] { 1.0, 8.0, null, 3.0, 4.0, null },
});
IUserSimilarity similarity = new PearsonCorrelationSimilarity(dataModel);
similarity.SetPreferenceInferrer(new TestPreferenceInferrer());
Assert.AreEqual(-0.435285750066007, similarity.UserSimilarity(1L, 2L), EPSILON);
}
public void testFromCorrelation()
{
IDataModel dataModel = getDataModel(
new long[] { 1, 2, 3 },
new Double?[][] {
new double?[] { 1.0, 2.0 },
new double?[] { 2.0, 5.0 },
new double?[] { 3.0, 6.0 },
});
IItemSimilarity otherSimilarity = new PearsonCorrelationSimilarity(dataModel);
IItemSimilarity itemSimilarity = new GenericItemSimilarity(otherSimilarity, dataModel);
assertCorrelationEquals(1.0, itemSimilarity.ItemSimilarity(0, 0));
assertCorrelationEquals(0.960768922830523, itemSimilarity.ItemSimilarity(0, 1));
}
public void testHowMany()
{
IDataModel dataModel = getDataModel(
new long[] {1, 2, 3, 4, 5},
new Double?[][] {
new double?[]{0.1, 0.2},
new double?[]{0.2, 0.3, 0.3, 0.6},
new double?[]{0.4, 0.4, 0.5, 0.9},
new double?[]{0.1, 0.4, 0.5, 0.8, 0.9, 1.0},
new double?[]{0.2, 0.3, 0.6, 0.7, 0.1, 0.2},
});
IUserSimilarity similarity = new PearsonCorrelationSimilarity(dataModel);
IUserNeighborhood neighborhood = new NearestNUserNeighborhood(2, similarity, dataModel);
IRecommender recommender = new GenericUserBasedRecommender(dataModel, neighborhood, similarity);
IList<IRecommendedItem> fewRecommended = recommender.Recommend(1, 2);
IList<IRecommendedItem> moreRecommended = recommender.Recommend(1, 4);
for (int i = 0; i < fewRecommended.Count; i++) {
Assert.AreEqual(fewRecommended[i].GetItemID(), moreRecommended[i].GetItemID());
}
recommender.Refresh(null);
for (int i = 0; i < fewRecommended.Count; i++) {
Assert.AreEqual(fewRecommended[i].GetItemID(), moreRecommended[i].GetItemID());
}
}
public void testSimpleItem()
{
IDataModel dataModel = getDataModel(
new long[] {1, 2, 3},
new Double?[][] {
new double?[]{1.0, 2.0},
new double?[]{2.0, 5.0},
new double?[]{3.0, 6.0},
});
double correlation =
new PearsonCorrelationSimilarity(dataModel).ItemSimilarity(0, 1);
assertCorrelationEquals(0.9607689228305227, correlation);
}
public void testSimpleItemWeighted()
{
IDataModel dataModel = getDataModel(
new long[] {1, 2, 3},
new Double?[][] {
new double?[]{1.0, 2.0},
new double?[]{2.0, 5.0},
new double?[]{3.0, 6.0},
});
IItemSimilarity itemSimilarity = new PearsonCorrelationSimilarity(dataModel, Weighting.WEIGHTED);
double correlation = itemSimilarity.ItemSimilarity(0, 1);
assertCorrelationEquals(0.9901922307076306, correlation);
}
public void testSimpleWeighted()
{
IDataModel dataModel = getDataModel(
new long[] {1, 2},
new Double?[][] {
new double?[]{1.0, 2.0, 3.0},
new double?[]{2.0, 5.0, 6.0},
});
double correlation = new PearsonCorrelationSimilarity(dataModel, Weighting.WEIGHTED).UserSimilarity(1, 2);
assertCorrelationEquals(0.9901922307076306, correlation);
}
public void testNoCorrelation1()
{
IDataModel dataModel = getDataModel(
new long[] {1, 2},
new Double?[][] {
new double?[]{3.0, -2.0},
new double?[]{-3.0, 2.0},
});
double correlation = new PearsonCorrelationSimilarity(dataModel).UserSimilarity(1, 2);
assertCorrelationEquals(-1.0, correlation);
}
public void testFullItemCorrelation1()
{
IDataModel dataModel = getDataModel(
new long[] {1, 2},
new Double?[][] {
new double?[]{3.0, 3.0},
new double?[]{-2.0, -2.0},
});
double correlation =
new PearsonCorrelationSimilarity(dataModel).ItemSimilarity(0, 1);
assertCorrelationEquals(1.0, correlation);
}
public void testNoItemCorrelation3()
{
IDataModel dataModel = getDataModel(
new long[] {1, 2, 3},
new Double?[][] {
new double?[]{90.0, 70.0},
new double?[]{80.0, 80.0},
new double?[]{70.0, 90.0},
});
double correlation =
new PearsonCorrelationSimilarity(dataModel).ItemSimilarity(0, 1);
assertCorrelationEquals(-1.0, correlation);
}
private static IUserBasedRecommender buildRecommender()
{
IDataModel dataModel = getDataModel();
IUserSimilarity similarity = new PearsonCorrelationSimilarity(dataModel);
IUserNeighborhood neighborhood = new NearestNUserNeighborhood(2, similarity, dataModel);
return new GenericUserBasedRecommender(dataModel, neighborhood, similarity);
}
public void testRescorer()
{
IDataModel dataModel = getDataModel(
new long[] {1, 2, 3},
new Double?[][] {
new double?[]{0.1, 0.2},
new double?[]{0.2, 0.3, 0.3, 0.6},
new double?[]{0.4, 0.5, 0.5, 0.9},
});
IUserSimilarity similarity = new PearsonCorrelationSimilarity(dataModel);
IUserNeighborhood neighborhood = new NearestNUserNeighborhood(2, similarity, dataModel);
IRecommender recommender = new GenericUserBasedRecommender(dataModel, neighborhood, similarity);
IList<IRecommendedItem> originalRecommended = recommender.Recommend(1, 2);
IList<IRecommendedItem> rescoredRecommended =
recommender.Recommend(1, 2, new ReversingRescorer<long>());
Assert.NotNull(originalRecommended);
Assert.NotNull(rescoredRecommended);
Assert.AreEqual(2, originalRecommended.Count);
Assert.AreEqual(2, rescoredRecommended.Count);
Assert.AreEqual(originalRecommended[0].GetItemID(), rescoredRecommended[1].GetItemID());
Assert.AreEqual(originalRecommended[1].GetItemID(), rescoredRecommended[0].GetItemID());
}
public void testInferrer()
{
IDataModel dataModel = getDataModel(
new long[] {1, 2},
new Double?[][] {
new double?[]{null, 1.0, 2.0, null, null, 6.0},
new double?[]{1.0, 8.0, null, 3.0, 4.0, null},
});
IUserSimilarity similarity = new PearsonCorrelationSimilarity(dataModel);
similarity.SetPreferenceInferrer(new TestPreferenceInferrer());
Assert.AreEqual(-0.435285750066007, similarity.UserSimilarity(1L, 2L), EPSILON);
}
public void testNoItemCorrelation2()
{
IDataModel dataModel = getDataModel(
new long[] {1, 2},
new Double?[][] {
new double?[]{null, 1.0, null},
new double?[]{null, null, 1.0},
});
double correlation =
new PearsonCorrelationSimilarity(dataModel).ItemSimilarity(1, 2);
Assert.True(Double.IsNaN(correlation));
}
public void testIsolatedUser()
{
IDataModel dataModel = getDataModel(
new long[] {1, 2, 3, 4},
new Double?[][] {
new double?[]{0.1, 0.2},
new double?[]{0.2, 0.3, 0.3, 0.6},
new double?[]{0.4, 0.4, 0.5, 0.9},
new double?[]{null, null, null, null, 1.0},
});
IUserSimilarity similarity = new PearsonCorrelationSimilarity(dataModel);
IUserNeighborhood neighborhood = new NearestNUserNeighborhood(3, similarity, dataModel);
IUserBasedRecommender recommender = new GenericUserBasedRecommender(dataModel, neighborhood, similarity);
long[] mostSimilar = recommender.MostSimilarUserIDs(4, 3);
Assert.NotNull(mostSimilar);
Assert.AreEqual(0, mostSimilar.Length);
}
public void testFile()
{
IUserSimilarity userSimilarity = new PearsonCorrelationSimilarity(model);
IUserNeighborhood neighborhood = new NearestNUserNeighborhood(3, userSimilarity, model);
IRecommender recommender = new GenericUserBasedRecommender(model, neighborhood, userSimilarity);
Assert.AreEqual(1, recommender.Recommend(123, 3).Count);
Assert.AreEqual(0, recommender.Recommend(234, 3).Count);
Assert.AreEqual(1, recommender.Recommend(345, 3).Count);
// Make sure this doesn't throw an exception
model.Refresh(null);
}
