net.librec.eval.ranking.HitRateEvaluator Maven / Gradle / Ivy
package net.librec.eval.ranking;
import net.librec.eval.AbstractRecommenderEvaluator;
import net.librec.math.structure.SparseMatrix;
import net.librec.recommender.item.ItemEntry;
import net.librec.recommender.item.RecommendedList;
import java.util.List;
/**
* HitRateEvaluator
*
* Xia Ning and George Karypis, SLIM: Sparse Linear Methods for Top-N Recommender Systems, ICDM 2011.
*
* They apply a leave-one-out validation method to evaluate the algorithm performance. In each run, each of the datasets
* is split into a training set and a testing set by randomly selecting one of the non-zero entries of each user and
* placing it into the testing set.
*
* @author WangYuFeng and Keqiang Wang
*/
public class HitRateEvaluator extends AbstractRecommenderEvaluator {
/**
* Evaluate on the test set with the the list of recommended items.
*
* @param testMatrix
* the given test set
* @param recommendedList
* the list of recommended items
* @return evaluate result
*/
public double evaluate(SparseMatrix testMatrix, RecommendedList recommendedList) {
if (testMatrix.size() == 0) {
return 0.0;
}
int totalHits = 0;
int numUsers = testMatrix.numRows();
int nonZeroNumUsers = 0;
for (int userID = 0; userID < numUsers; userID++) {
List testListByUser = testMatrix.getColumns(userID);
if (testListByUser.size() == 1) {
int itemIdx = testListByUser.get(0);
List> recommendListByUser = recommendedList.getItemIdxListByUserIdx(userID);
int topK = this.topN <= recommendListByUser.size() ? this.topN : recommendListByUser.size();
for (int indexOfItem = 0; indexOfItem < topK; indexOfItem++) {
int recommendItemIdx = recommendListByUser.get(indexOfItem).getKey();
if (recommendItemIdx == itemIdx) {
totalHits++;
break;
}
}
nonZeroNumUsers++;
} else if (testListByUser.size() > 1) {
throw new IndexOutOfBoundsException("It is not a leave-one-out validation method! Please use leave-one-out validation method");
}
}
return nonZeroNumUsers > 0 ? 1.0 * totalHits / nonZeroNumUsers : 0.0d;
}
}