If you are using one of the standard RecommenderEvaluator implementations, then this will not happen. It already splits the data into 'test' and 'training' data. It will not try to estimate a preference that is already known in the training data, no.
On Thu, Nov 19, 2009 at 5:40 PM, jamborta <[email protected]> wrote: > > sorry, I have one more thing to add, in GenericUserBasedRecommender this is > the method that estimates the preference: > > public float estimatePreference(long userID, long itemID) throws > TasteException { > DataModel model = getDataModel(); > Float actualPref = model.getPreferenceValue(userID, itemID); > if (actualPref != null) { > return actualPref; > } > long[] theNeighborhood = neighborhood.getUserNeighborhood(userID); > return doEstimatePreference(userID, theNeighborhood, itemID); > } > > but I try to estimate from the training set, so obviously the userID and > itemID exsist in the model, therefore it returns the actualPref. that's why > I get 0 as a result. > -- > View this message in context: > http://old.nabble.com/evaluating-recommender-systems-tp26421408p26421411.html > Sent from the Mahout User List mailing list archive at Nabble.com. > >
