/lenskit-eval-balke/src/integration-test/matthias-balke-da.groovy

/*
 * LensKit, an open source recommender systems toolkit.
 * Copyright 2010-2011 Regents of the University of Minnesota
 *
 * This program is free software; you can redistribute it and/or modify
 * it under the terms of the GNU Lesser General Public License as
 * published by the Free Software Foundation; either version 2.1 of the
 * License, or (at your option) any later version.
 *
 * This program is distributed in the hope that it will be useful, but WITHOUT
 * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS
 * FOR A PARTICULAR PURPOSE. See the GNU General Public License for more
 * details.
 *
 * You should have received a copy of the GNU General Public License along with
 * this program; if not, write to the Free Software Foundation, Inc., 51
 * Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
 */
import org.grouplens.lenskit.ItemRecommender
import org.grouplens.lenskit.RatingPredictor
import org.grouplens.lenskit.baseline.BaselinePredictor
import org.grouplens.lenskit.baseline.GlobalMeanPredictor
import org.grouplens.lenskit.baseline.ItemMeanPredictor
import org.grouplens.lenskit.baseline.ItemUserMeanPredictor
import org.grouplens.lenskit.baseline.UserMeanPredictor
import org.grouplens.lenskit.eval.data.crossfold.RandomOrder
import org.grouplens.lenskit.eval.metrics.StatisticsMetric
import org.grouplens.lenskit.eval.metrics.predict.CoveragePredictMetric
import org.grouplens.lenskit.eval.metrics.predict.MAEPredictMetric;
import org.grouplens.lenskit.eval.metrics.predict.RMSEPredictMetric
import org.grouplens.lenskit.eval.metrics.recommend.DiversityRecommendMetric
import org.grouplens.lenskit.eval.metrics.recommend.FScoreRecommendMetric
import org.grouplens.lenskit.eval.metrics.recommend.DiversityRecommendMetric.DiversityType
import org.grouplens.lenskit.knn.item.ItemItemRatingPredictor
import org.grouplens.lenskit.knn.item.ItemItemRecommender;
import org.grouplens.lenskit.knn.params.NeighborhoodSize
import org.grouplens.lenskit.knn.params.SimilarityDamping
import org.grouplens.lenskit.norm.BaselineSubtractingNormalizer
import org.grouplens.lenskit.norm.VectorNormalizer
import org.grouplens.lenskit.params.NormalizerBaseline
import org.grouplens.lenskit.params.UserVectorNormalizer
import org.grouplens.lenskit.rfrec.RFRecRatingPredictor
import org.grouplens.lenskit.rfrec.RFRecRecommender
import org.grouplens.lenskit.rfrec.parameters.DefaultItemWeight
import org.grouplens.lenskit.rfrec.parameters.DefaultUserWeight
import org.grouplens.lenskit.rfrec.parameters.Gamma
import org.grouplens.lenskit.rfrec.parameters.Iterations
import org.grouplens.lenskit.rfrec.parameters.Lambda
import org.grouplens.lenskit.slopeone.SlopeOneRecommender;
import org.grouplens.lenskit.slopeone.WeightedSlopeOneRatingPredictor
import org.grouplens.lenskit.slopeone.params.DeviationDamping
import org.grouplens.lenskit.svd.FunkSVDRatingPredictor
import org.grouplens.lenskit.svd.FunkSVDRecommender;
import org.grouplens.lenskit.svd.params.FeatureCount
import org.grouplens.lenskit.svd.params.IterationCount

def buildDir = System.getProperty("project.build.directory", ".")

def ml1M = crossfold("ml1M") {
    source csvfile("${buildDir}/ml-1M/ratings.dat") {
        delimiter "::"
        domain {
            minimum 1.0
            maximum 5.0
            precision 1.0
        }
    }
    order RandomOrder
    holdout 10 
    partitions 5
    train "${buildDir}/ml-1M/ml-1M.train.%d.csv"
    test "${buildDir}/ml-1M/ml-1M.test.%d.csv"
}

def netflix1M = crossfold("Netflix1M") {
    source csvfile("${buildDir}/Netflix-1M/Netflix-1M-dense.mldata.gz") {
        delimiter ","
        domain {
            minimum 1.0
            maximum 5.0
            precision 1.0
        }
    }
    order RandomOrder
    holdout 10
    partitions 5
    train "${buildDir}/Netflix-1M/Netflix-1M.train.%d.csv"
    test "${buildDir}/Netflix-1M/Netflix-1M.test.%d.csv"
}

def yahooMovies200k = crossfold("YahooMovies200k") {
    source csvfile("${buildDir}/YahooMovies-200k/YahooMovies-200k.mldata.gz") {
        delimiter ","
        domain {
            minimum 1.0
            maximum 13.0
            precision 1.0
        }
    }
    order RandomOrder
    holdout 0.3
    partitions 5
    train "${buildDir}/YahooMovies-200k/YahooMovies-200k.train.%d.csv"
    test "${buildDir}/YahooMovies-200k/YahooMovies-200k.test.%d.csv"
}

trainTest {
    depends ml1M
    dataset ml1M

    depends netflix1M
    dataset netflix1M

    depends yahooMovies200k
    dataset yahooMovies200k

    output "${buildDir}/eval-output/eval-output.csv"

    numRecs 5

    metric CoveragePredictMetric
    metric MAEPredictMetric
    metric RMSEPredictMetric

    metric new FScoreRecommendMetric()

    metric new DiversityRecommendMetric(DiversityType.HITS)
    metric new DiversityRecommendMetric(DiversityType.POPULARITY)

    metric new StatisticsMetric()

    //    algorithm("UserUser") {
    //        setComponent(RatingPredictor, UserUserRatingPredictor)
    //        setComponent(PredictNormalizer, VectorNormalizer, MeanVarianceNormalizer)
    //        setComponent(BaselinePredictor, ItemUserMeanPredictor)
    //        setComponent(NormalizerBaseline, BaselinePredictor, UserMeanPredictor)
    //        setComponent(UserVectorNormalizer, VectorNormalizer, BaselineSubtractingNormalizer)
    //        setComponent(UserSimilarity, Similarity, CosineSimilarity)
    //        set(SimilarityDamping, 100)
    //        set(NeighborhoodSize, 30)
    //    }
    //

    algorithm("ItemItem") {
        setComponent(RatingPredictor, ItemItemRatingPredictor)
        setComponent(ItemRecommender, ItemItemRecommender)
        setComponent(BaselinePredictor, ItemUserMeanPredictor)
        setComponent(UserVectorNormalizer, VectorNormalizer, BaselineSubtractingNormalizer)
        set(SimilarityDamping, 100)
        set(NeighborhoodSize, 30);
    }

    algorithm("WeightedSlopeOne") {
        setComponent(RatingPredictor, WeightedSlopeOneRatingPredictor)
        setComponent(ItemRecommender, SlopeOneRecommender)
        setComponent(NormalizerBaseline, BaselinePredictor, GlobalMeanPredictor)
        setComponent(UserVectorNormalizer, VectorNormalizer, BaselineSubtractingNormalizer)
        setComponent(BaselinePredictor, ItemUserMeanPredictor)
        set(DeviationDamping, 0)
    }

    algorithm("RFRec"){
        setComponent(ItemRecommender, RFRecRecommender)
        setComponent(RatingPredictor, RFRecRatingPredictor)
        set(DefaultUserWeight, 0.4)
        set(DefaultItemWeight, 0.6)
        set(Iterations, 15)
        set(Gamma, 0.001)
        set(Lambda, 0.002)
    }

    algorithm("FunkSVD") {
        setComponent(RatingPredictor, FunkSVDRatingPredictor)
        setComponent(ItemRecommender, FunkSVDRecommender)
        setComponent(BaselinePredictor, ItemUserMeanPredictor)
        set(FeatureCount, 30)
        set(IterationCount, 100)
    }

}