/**
* {@inheritDoc}
*/
public double Evaluate(RecommenderBuilder recommenderBuilder,
DataModel dataModel,
double trainingPercentage,
double evaluationPercentage)
{
if (recommenderBuilder == null)
{
throw new ArgumentNullException("recommenderBuilder is null");
}
if (dataModel == null)
{
throw new ArgumentNullException("dataModel is null");
}
if (double.IsNaN(trainingPercentage) || trainingPercentage <= 0.0 || trainingPercentage >= 1.0)
{
throw new ArgumentException("Invalid trainingPercentage: " + trainingPercentage);
}
if (double.IsNaN(evaluationPercentage) || evaluationPercentage <= 0.0 || evaluationPercentage > 1.0)
{
throw new ArgumentException("Invalid evaluationPercentage: " + evaluationPercentage);
}
log.Info("Beginning evaluation using " + trainingPercentage + " of " + dataModel);
int numUsers = dataModel.GetNumUsers();
ICollection<User> trainingUsers = new List<User>(1 + (int) (trainingPercentage * (double) numUsers));
IDictionary<User, ICollection<Preference>> testUserPrefs =
new Dictionary<User, ICollection<Preference>>(1 + (int) ((1.0 - trainingPercentage) * (double) numUsers));
foreach (User user in dataModel.GetUsers())
{
if (random.NextDouble() < evaluationPercentage)
{
ICollection<Preference> trainingPrefs = new List<Preference>();
ICollection<Preference> testPrefs = new List<Preference>();
Preference[] prefs = user.GetPreferencesAsArray();
foreach (Preference pref in prefs)
{
Item itemCopy = new GenericItem<String>(pref.Item.ID.ToString());
Preference newPref = new GenericPreference(null, itemCopy, pref.Value);
if (random.NextDouble() < trainingPercentage)
{
trainingPrefs.Add(newPref);
} else {
testPrefs.Add(newPref);
}
}
if (log.IsDebugEnabled) {
log.Debug("Training against " + trainingPrefs.Count + " preferences");
log.Debug("Evaluating accuracy of " + testPrefs.Count + " preferences");
}
if (trainingPrefs.Count > 0)
{
User trainingUser = new GenericUser<String>(user.ID.ToString(), trainingPrefs);
trainingUsers.Add(trainingUser);
if (testPrefs.Count > 0)
{
testUserPrefs.Add(trainingUser, testPrefs);
}
}
}
}
DataModel trainingModel = new GenericDataModel(trainingUsers);
Recommender recommender = recommenderBuilder.BuildRecommender(trainingModel);
double result = GetEvaluation(testUserPrefs, recommender);
log.Info("Evaluation result: " + result);
return result;
}
public Preference[] GetPreferencesAsArray()
{
Preference[] sortedPrefs = delegateUser.GetPreferencesAsArray();
Array.Sort(sortedPrefs, ByValuePreferenceComparer.Instance);
for (int i = 0; i < sortedPrefs.Length; i++)
{
sortedPrefs[i] = new GenericPreference(this, sortedPrefs[i].Item, (double) (i + 1));
}
Array.Sort(sortedPrefs, ByItemPreferenceComparer.Instance);
return sortedPrefs;
}
public void testNoCorrelation2()
{
Preference pref1 = new GenericPreference(null, new GenericItem<String>("1"), 1.0);
GenericUser<String> user1 = new GenericUser<String>("test1", ScalarToList<Preference>(pref1));
Preference pref2 = new GenericPreference(null, new GenericItem<String>("2"), 1.0);
GenericUser<String> user2 = new GenericUser<String>("test2", ScalarToList<Preference>(pref2));
DataModel dataModel = GetDataModel(user1, user2);
double correlation = new PearsonCorrelation(dataModel).GetUserCorrelation(user1, user2);
Assert.IsTrue(Double.IsNaN(correlation));
}
public IEnumerable<Preference> GetPreferences()
{
// todo: cache this
Preference[] source = delegateUser.GetPreferencesAsArray();
int length = source.Length;
Preference[] sortedPrefs = new Preference[length];
Array.Copy(source, sortedPrefs, length);
for (int i = 0; i < length; i++)
{
Item item = sortedPrefs[i].Item;
sortedPrefs[i] = new GenericPreference(this, item, (double) (i + 1));
}
Array.Sort<Preference>(sortedPrefs, ByValuePreferenceComparer.Instance);
return sortedPrefs;
}
