func capper(v float64, m model.DataModel) float64 {
if v > m.MaxPreferenceValue() {
v = m.MaxPreferenceValue()
} else if v < m.MinPreferenceValue() {
v = m.MinPreferenceValue()
}
return v
}
func (this *GenericItemBasedRecommender) getAllOtherItemIds(preferredItemIds []uint64, dataModel model.DataModel) ([]uint64, error) {
possibleIdSet := utils.IdSet{}
for _, itemId := range preferredItemIds {
itemPrefs, err := dataModel.GetItemPreferences(itemId)
if err != nil {
continue
}
for _, uid := range itemPrefs.Ids() {
up, err := dataModel.GetUserPreferences(uid)
if err != nil {
continue
}
possibleIdSet.AddArray(up.Ids())
}
}
possibleIdSet.RemoveArray(preferredItemIds)
return possibleIdSet.ToArray(), nil
}
func SplitTrainingAndTest(m model.DataModel, trainingPercentage, evaluationPercentage float32) (model.PreferenceArrayMap, model.PreferenceArrayMap) {
trainings := model.NewUserPreferenceArrayMap()
tests := model.NewUserPreferenceArrayMap()
for _, uid := range m.UserIds() {
if rand.Float32() < evaluationPercentage {
prefs, err := m.GetUserPreferences(uid)
if err != nil {
continue
}
for _, p := range prefs.Raw() {
if rand.Float32() < trainingPercentage {
trainings.Set(p)
} else {
tests.Set(p)
}
}
}
}
return trainings, tests
}
