yes I get zero. I've just figured that if I do it this way, it works fine:
public class ItemBasedBuilder implements RecommenderBuilder {
@Override
public Recommender buildRecommender(DataModel dataModel) throws
TasteException {
UserSimilarity userSimilarity = new
PearsonCorrelationSimilarity(dataModel);
UserNeighborhood neighborhood =
new NearestNUserNeighborhood(10, userSimilarity, dataModel);
Recommender recommender =
new GenericUserBasedRecommender(dataModel, neighborhood,
userSimilarity);
return recommender;
}
}
but these two solution should give the same result I guess...
srowen wrote:
>
> This looks fine to me (though you are building a user-based
> recommender, and from the sound of it, you intended to build an
> item-based one). You are getting an evaluation result of 0?
>
> Are you using the latest code (version 0.2)?
>
> Also, try increasing the amount of data you use. Your "0.8" can be
> "0.95" to train on 95% of the data. Also, for a small data set like
> this, pass 1.0 as the last parameter to use all of it.
>
> At the moment you're training on 16,000 ratings, which should still
> give some non-trivial result, but that's not a lot.
>
> On Fri, Nov 20, 2009 at 11:15 AM, jamborta <[email protected]> wrote:
>>
>> thanks, i'm using this one with the standard movielens (100k) dataset.
>>
>> public Recommender buildRecommender(DataModel dataModel) throws
>> TasteException {
>> DataModel model = null;
>> try {
>> model = new FileDataModel(new File("./data/all_data.data"));
>>
>> } catch (FileNotFoundException e) {
>> e.printStackTrace();
>> }
>> UserSimilarity userSimilarity = new
>> PearsonCorrelationSimilarity(model);
>> UserNeighborhood neighborhood =
>> new NearestNUserNeighborhood(10, userSimilarity, model);
>> Recommender recommender =
>> new GenericUserBasedRecommender(dataModel, neighborhood,
>> userSimilarity);
>> return recommender;
>>
>> }
>>
>>
>> srowen wrote:
>>>
>>> 0 is very good! But I agree, it is probably an error.
>>>
>>> I see you call evaluate() twice. This is not necessary. Call it once,
>>> and save the result, then print it. But this is not the issue.
>>>
>>> What is in the ItemBasedBuilder class? what is your data like? Maybe
>>> if I can see this I can suggest why you get this result.
>>>
>>> On Thu, Nov 19, 2009 at 5:31 PM, jamborta <[email protected]> wrote:
>>>>
>>>> hi.
>>>>
>>>> i'm not sure if this is a bug or I do somthing wrong, but when I try to
>>>> evaluate a system it returns 0 as a result. I'm using this piece of
>>>> code:
>>>>
>>>> DataModel model = new FileDataModel(new
>>>> File("./data/all_data.data"));
>>>> RecommenderBuilder build = new ItemBasedBuilder();
>>>> AverageAbsoluteDifferenceRecommenderEvaluator evaluate = new
>>>> AverageAbsoluteDifferenceRecommenderEvaluator();
>>>> DataModelBuilder model2 = null;
>>>> evaluate.evaluate(build, model2, model,0.8,0.2 );
>>>> System.out.println(evaluate.evaluate(build, model2,
>>>> model,0.8,0.2 ));
>>>>
>>>> thanks a lot.
>>>> --
>>>> View this message in context:
>>>> http://old.nabble.com/evaluating-recommender-systems-tp26421408p26421408.html
>>>> Sent from the Mahout User List mailing list archive at Nabble.com.
>>>>
>>>>
>>>
>>>
>>
>> --
>> View this message in context:
>> http://old.nabble.com/evaluating-recommender-systems-tp26421408p26438752.html
>> Sent from the Mahout User List mailing list archive at Nabble.com.
>>
>>
>
>
--
View this message in context:
http://old.nabble.com/evaluating-recommender-systems-tp26421408p26443610.html
Sent from the Mahout User List mailing list archive at Nabble.com.
