package org.mahout.recommendations;
import org.apache.mahout.cf.taste.common.TasteException;
import org.apache.mahout.cf.taste.eval.RecommenderBuilder;
import org.apache.mahout.cf.taste.eval.RecommenderEvaluator;
import org.apache.mahout.cf.taste.impl.eval.AverageAbsoluteDifferenceRecommenderEvaluator;
import org.apache.mahout.cf.taste.impl.model.file.FileDataModel;
import org.apache.mahout.cf.taste.impl.neighborhood.NearestNUserNeighborhood;
import org.apache.mahout.cf.taste.impl.recommender.GenericUserBasedRecommender;
import org.apache.mahout.cf.taste.impl.similarity.PearsonCorrelationSimilarity;
import org.apache.mahout.cf.taste.model.DataModel;
import org.apache.mahout.cf.taste.neighborhood.UserNeighborhood;
import org.apache.mahout.cf.taste.recommender.Recommender;
import org.apache.mahout.cf.taste.similarity.UserSimilarity;
import org.apache.mahout.common.RandomUtils;
import java.io.File;
/**
* @author Krisztian Horvath
*/
public class RecommenderIntroEvaluator {
public static void main(String[] args) throws Exception {
RandomUtils.useTestSeed();
DataModel model = new FileDataModel(new File("src/main/resources/intro.csv"));
RecommenderEvaluator evaluator = new AverageAbsoluteDifferenceRecommenderEvaluator();
RecommenderBuilder builder = new RecommenderBuilder() {
@Override
public Recommender buildRecommender(DataModel model) throws TasteException {
UserSimilarity similarity = new PearsonCorrelationSimilarity(model);
UserNeighborhood neighborhood = new NearestNUserNeighborhood(2, similarity, model);
return new GenericUserBasedRecommender(model, neighborhood, similarity);
}
};
double score = evaluator.evaluate(builder, null, model, 0.8, 1.0);
System.out.println(score);
}
}