public void TestMovieLensSingle()
{
// step 1: dataset
var config = new CsvConfiguration()
{
Delimiter = "::",
HasHeaderRecord = true
};
// load data
var trainReader = new CsvReader(Paths.MovieLens1MTrain75, config);
var testReader = new CsvReader(Paths.MovieLens1MTest25, config, true);
var container = new DataContainer();
trainReader.LoadData(container);
testReader.LoadData(container);
var startTime = DateTime.Now;
var splitter = new RatingSimpleSplitter(container);
//var recommender = new MediaLiteRatingPredictor(new MatrixFactorization());
var recommender = new LibFmTrainTester(libFmPath: "LibFm.Net.64.exe");
// evaluation
var ctx = new EvalutationContext<ItemRating>(recommender, splitter);
var ep = new EvaluationPipeline<ItemRating>(ctx);
ep.Evaluators.Add(new RMSE());
ep.Run();
var duration = (int)DateTime.Now.Subtract(startTime).TotalMilliseconds;
Console.WriteLine("RMSE\tDuration\n{0}\t{1}", ctx["RMSE"], duration);
}
public void TestImplicitTrust()
{
var socialReguls = new float[] { 1F };
var numFactors = new uint[] { 5, 10 };
var trustScores = new string[] { "trust_values_LATHIA.dat", "trust_values_HWANGCHEN.dat",
"trust_values_ODONOVAN.dat", "trust_values_PEARSON.dat", "trust_values_SHAMBOURLU.dat" };
// step 1: dataset
var config = new CsvConfiguration();
config.Delimiter = " ";
var trainReader = new CsvReader <ItemRating>(Paths.EpinionTrain80, config, new ItemRatingMap());
var testReader = new CsvReader <ItemRating>(Paths.EpinionTest20, config, new ItemRatingMap());
var dataset = new Dataset <ItemRating>(trainReader, testReader);
foreach (string scoreFile in trustScores)
{
var relations = File.ReadAllLines(Paths.EpinionRelationsImplicit + scoreFile).ToCsvDictionary('\t')
.Select(i => new Relation()
{
UserId = i["UserId"],
ConnectedId = i["ConnectionId"],
ConnectionStrength = float.Parse(i["Strength"])
});
//.Where(r => r.ConnectionStrength > 1F);
string rmseValues = "", maeValues = "";
foreach (uint num in numFactors)
{
// step 2: recommender
var algorithm = new SocialMF();
algorithm.SocialRegularization = 1;
algorithm.NumFactors = num;
var recommender = new MediaLiteRatingPredictor(algorithm, relations);
// step3: evaluation
var context = new EvalutationContext <ItemRating>(recommender, dataset);
var ep = new EvaluationPipeline <ItemRating>(context);
ep.Evaluators.Add(new RMSE());
ep.Evaluators.Add(new MAE());
ep.Run();
rmseValues += context["RMSE"] + "\t";
maeValues += context["MAE"] + "\t";
}
Console.WriteLine(scoreFile + "\t" + rmseValues + "\t" + maeValues);
}
}
public void TestExplicitTrust()
{
//var socialReguls = new float[] { 0.1F, 0.2F, 0.5F, 0.8F, 1F, 1.5F, 2F, 3F, 5F};
//var numFactors = new uint[] {2, 5, 10, 15, 20};
var socialReguls = new float[] { 1 };
var numFactors = new uint[] { 5, 10 };
// step 1: dataset
var config = new CsvConfiguration();
config.Delimiter = " ";
var trainReader = new CsvReader <ItemRating>(Paths.EpinionTrain75, config, new ItemRatingMap());
var testReader = new CsvReader <ItemRating>(Paths.EpinionTest25, config, new ItemRatingMap());
var dataset = new Dataset <ItemRating>(trainReader, testReader);
var relations = File.ReadAllLines(Paths.EpinionRelations).ToCsvDictionary(' ')
.Select(i => new Relation()
{
UserId = i["UserId"], ConnectedId = i["ConnectionId"], DatasetId = 1
});
foreach (uint num in numFactors)
{
string rmseValues = "", maeValues = "";
foreach (float regul in socialReguls)
{
// step 2: recommender
var algorithm = new MatrixFactorization();
algorithm.NumFactors = num;
//algorithm.SocialRegularization = regul;
var recommender = new MediaLiteRatingPredictor(algorithm, relations);
// step3: evaluation
var context = new EvalutationContext <ItemRating>(recommender, dataset);
var ep = new EvaluationPipeline <ItemRating>(context);
ep.Evaluators.Add(new RMSE());
ep.Evaluators.Add(new MAE());
ep.Run();
rmseValues += context["RMSE"] + "\t";
maeValues += context["MAE"] + "\t";
}
Console.WriteLine(num + "\t" + rmseValues + "\t" + maeValues);
}
}
public void Evaluate(EvalutationContext <ItemRating> context)
{
// make sure that the test samples are predicted
context.RunDefaultTrainAndTest();
var output = context.Dataset.TestSamples.AsEnumerable()
.Select(ir =>
{
var t = _container.Tweets[ir];
return(string.Format("{0},{1},{2}", t.TwitterUserId, t.Id, ir.PredictedRating));
});
File.WriteAllLines(_outputFile, output);
}
public void TestMovieLensWithClusters()
{
// Todo: instead of loading the dataset every time add a updateCluster method to the Dataset<ItemRatingWithCluster>
List <string> rmses = new List <string>();
List <string> maes = new List <string>();
for (int i = 0; i < 15; i += 2)
{
string usersClusterFile = Paths.MovieLens1MUsersCluster + i + ".csv";
string itemsClusterFile = Paths.MovieLens1MItemsCluster + i + ".csv";
MovieLensReader trainReader, testReader;
if (i == 0)
{
trainReader = new MovieLensReader(Paths.MovieLens1MTrain75);
testReader = new MovieLensReader(Paths.MovieLens1MTest25);
}
else
{
trainReader = new MovieLensReader(Paths.MovieLens1MTrain75, usersClusterFile, itemsClusterFile);
testReader = new MovieLensReader(Paths.MovieLens1MTest25, usersClusterFile, itemsClusterFile);
}
var dataset = new Dataset <MovieLensItemRating>(trainReader, testReader);
var recommender = new LibFmTrainTester();
var context = new EvalutationContext <ItemRating>(recommender, dataset);
var ep = new EvaluationPipeline <ItemRating>(context);
ep.Evaluators.Add(new RMSE());
ep.Evaluators.Add(new MAE());
ep.Run();
rmses.Add(context["RMSE"].ToString());
maes.Add(context["MAE"].ToString());
}
Console.WriteLine("RMSEs--------------");
rmses.ForEach(Console.WriteLine);
Console.WriteLine("MAEs-------------");
maes.ForEach(Console.WriteLine);
}
public void TestEpinionClusters()
{
List <string> rmses = new List <string>();
List <string> maes = new List <string>();
for (int i = 0; i < 15; i += 2)
{
string usersClusterFile = Paths.EpinionUsersCluster + i + ".csv";
string itemsClusterFile = Paths.EpinionItemsCluster + i + ".csv";
EpinionReader trainReader, testReader;
if (i == 0)
{
trainReader = new EpinionReader(Paths.EpinionTrain75);
testReader = new EpinionReader(Paths.EpinionTest25);
}
else
{
trainReader = new EpinionReader(Paths.EpinionTrain75, usersClusterFile, itemsClusterFile);
testReader = new EpinionReader(Paths.EpinionTest25, usersClusterFile, itemsClusterFile);
}
var dataset = new Dataset <EpinionItemRating>(trainReader, testReader);
var recommender = new LibFmTrainTester();
var context = new EvalutationContext <ItemRating>(recommender, dataset);
var ep = new EvaluationPipeline <ItemRating>(context);
ep.Evaluators.Add(new RMSE());
ep.Evaluators.Add(new MAE());
ep.Run();
rmses.Add(context["RMSE"].ToString());
maes.Add(context["MAE"].ToString());
}
Console.WriteLine("RMSEs--------------");
rmses.ForEach(Console.WriteLine);
Console.WriteLine("MAEs-------------");
maes.ForEach(Console.WriteLine);
}
public void TestAmazonWithNmfCluster()
{
var trainReader = new AmazonReader(Paths.AmazonBooksTrain75, Paths.AmazonBooksUsersCluster + ".nmf.u", Paths.AmazonBooksUsersCluster + ".nmf.i");
var testReader = new AmazonReader(Paths.AmazonBooksTest25, Paths.AmazonBooksUsersCluster + ".nmf.u", Paths.AmazonBooksUsersCluster + ".nmf.i");
var dataset = new Dataset <ItemRatingWithClusters>(trainReader, testReader);
var recommender = new LibFmTrainTester();
var context = new EvalutationContext <ItemRating>(recommender, dataset);
var ep = new EvaluationPipeline <ItemRating>(context);
ep.Evaluators.Add(new RMSE());
ep.Evaluators.Add(new MAE());
ep.Run();
}
public void TestEpinionsTrustAware()
{
// step 1: dataset
var config = new CsvConfiguration()
{
Delimiter = " ",
HasHeaderRecord = true
};
// load data
var trainReader = new CsvReader(Paths.EpinionTrain80, config);
var testReader = new CsvReader(Paths.EpinionTest20, config, true);
var readers = new List <IDatasetReader>()
{
trainReader, testReader
};
var epinionTrustReader = new EpinionTrustReader(readers.ToArray(), Paths.EpinionRelationsImplicit);
var container = new DataContainer();
epinionTrustReader.LoadData(container);
var startTime = DateTime.Now;
var splitter = new RatingSimpleSplitter(container);
//var recommender = new MediaLiteRatingPredictor(new MatrixFactorization());
var fb = new TrustAwareLibFmFeatureBuilder(container, 4, true);
var recommender = new LibFmTrainTester(featureBuilder: fb);
// evaluation
var ctx = new EvalutationContext <ItemRating>(recommender, splitter);
var ep = new EvaluationPipeline <ItemRating>(ctx);
ep.Evaluators.Add(new RMSE());
ep.Run();
var duration = (int)DateTime.Now.Subtract(startTime).TotalMilliseconds;
Console.WriteLine("RMSE\tDuration\n{0}\t{1}", ctx["RMSE"], duration);
}
public void TestMovieLensSingleDomain()
{
int numDomains = 1;
// load data
var movieLensReader = new MovieLensCrossDomainReader(Paths.MovieLens1MMovies, Paths.MovieLens1M);
var container = new MovieLensCrossDomainContainer(numDomains);
movieLensReader.LoadData(container);
// set taget and active domains
var targetDomain = container.SpecifyTargetDomain("ml0");
container.PrintStatistics();
var startTime = DateTime.Now;
var splitter = new CrossDomainSimpleSplitter(container, 0.25f);
// recommender with non-CrossDomain feature builder
var model = new MatrixFactorization();
model.NumIter = 50;
model.NumFactors = 8;
model.Regularization = 0.1f;
//var recommender = new MediaLiteRatingPredictor(model);
var recommender = new LibFmTrainTester();
// evaluation
var ctx = new EvalutationContext <ItemRating>(recommender, splitter);
var ep = new EvaluationPipeline <ItemRating>(ctx);
ep.Evaluators.Add(new RMSE());
ep.Evaluators.Add(new MAE());
ep.Run();
var duration = (int)DateTime.Now.Subtract(startTime).TotalMilliseconds;
Console.WriteLine("RMSE\tDuration\n{0}\t{1}", ctx["RMSE"], duration);
//container.CreateClusterFiles(Paths.MovieLens1M.GetDirectoryPath() + "\\movies.clust.raw", Paths.MovieLens1M.GetDirectoryPath() + "\\movies.clust.feat");
//container.WriteClusters(Paths.MovieLens1M.GetDirectoryPath() + "\\movies.clust");
}
static void TrainTestFm()
{
var reader = new PlayingSessionReader(path);
var container = new MusicDataContainer();
reader.LoadData(container);
var splitter = new RatingSimpleSplitter(container, 0.25f);
//var itemRecommender = new MostPopular();
//var model = new MediaLitePosFeedbakItemRecommender(itemRecommender);
var fm = new LibFmTrainTester();
var context = new EvalutationContext<ItemRating>(fm, splitter);
var pipline = new EvaluationPipeline<ItemRating>(context);
pipline.Evaluators.Add(new RMSE());
pipline.Evaluators.Add(new MAE());
pipline.Run();
}
static void TrainTestFm()
{
var reader = new PlayingSessionReader(path);
var container = new MusicDataContainer();
reader.LoadData(container);
var splitter = new RatingSimpleSplitter(container, 0.25f);
//var itemRecommender = new MostPopular();
//var model = new MediaLitePosFeedbakItemRecommender(itemRecommender);
var fm = new LibFmTrainTester();
var context = new EvalutationContext <ItemRating>(fm, splitter);
var pipline = new EvaluationPipeline <ItemRating>(context);
pipline.Evaluators.Add(new RMSE());
pipline.Evaluators.Add(new MAE());
pipline.Run();
}
static void TrainTest()
{
var reader = new PlayingSessionReader(path);
var container = new MusicDataContainer();
reader.LoadData(container);
var splitter = new PositiveFeedbackSimpleSplitter(container, 0.3f);
//var itemRecommender = new MostPopular();
//var model = new MediaLitePosFeedbakItemRecommender(itemRecommender);
var fm = new PosFeedbackLibFmTrainTester();
var context = new EvalutationContext <PositiveFeedback>(fm, splitter);
//var pipline = new EvaluationPipeline<PositiveFeedback>(context);
//pipline.Evaluators.Add(new MediaLitePositiveFeedbackEvaluators(itemRecommender));
//pipline.Run();
context.RunDefaultTrainAndTest();
}
static void TrainTest()
{
var reader = new PlayingSessionReader(path);
var container = new MusicDataContainer();
reader.LoadData(container);
var splitter = new PositiveFeedbackSimpleSplitter(container, 0.3f);
//var itemRecommender = new MostPopular();
//var model = new MediaLitePosFeedbakItemRecommender(itemRecommender);
var fm = new PosFeedbackLibFmTrainTester();
var context = new EvalutationContext<PositiveFeedback>(fm, splitter);
//var pipline = new EvaluationPipeline<PositiveFeedback>(context);
//pipline.Evaluators.Add(new MediaLitePositiveFeedbackEvaluators(itemRecommender));
//pipline.Run();
context.RunDefaultTrainAndTest();
}
public void TestAmazonAllDomains2()
{
// step 1: dataset
var config = new CsvConfiguration()
{
Delimiter = ",",
HasHeaderRecord = true
};
var container = new DataContainer();
var bookReader = new CsvReader(Paths.AmazonBooksRatings, config);
var musicReader = new CsvReader(Paths.AmazonMusicRatings, config);
var dvdReader = new CsvReader(Paths.AmazonDvdRatings, config);
var videoReader = new CsvReader(Paths.AmazonVideoRatings, config);
bookReader.LoadData(container);
musicReader.LoadData(container);
dvdReader.LoadData(container);
videoReader.LoadData(container);
var splitter = new RatingSimpleSplitter(container, 0.25f);
var startTime = DateTime.Now;
var recommender = new LibFmTrainTester();
//var recommender = new MediaLiteRatingPredictor(new MatrixFactorization());
// step3: evaluation
var ctx = new EvalutationContext <ItemRating>(recommender, splitter);
var ep = new EvaluationPipeline <ItemRating>(ctx);
ep.Evaluators.Add(new RMSE());
ep.Evaluators.Add(new MAE());
ep.Run();
var duration = DateTime.Now.Subtract(startTime);
Console.WriteLine("RMSE\t{0}\nDuration\t{1}", ctx["RMSE"], (int)duration.TotalMilliseconds);
}
public void TestTrustWithFM()
{
// step 1: dataset
var config = new CsvConfiguration();
config.Delimiter = " ";
var trainReader = new CsvReader <ItemRating>(Paths.EpinionTrain80, config, new ItemRatingMap());
var testReader = new CsvReader <ItemRating>(Paths.EpinionTest20, config, new ItemRatingMap());
var relations = File.ReadAllLines(Paths.EpinionRelationsImplicit).ToCsvDictionary('\t')
.Select(i => new Relation()
{
UserId = i["UserId"],
ConnectedId = i["ConnectionId"],
ConnectionStrength = float.Parse(i["Strength"])
}).ToList();
//.Where(r => r.ConnectionStrength > 1F);
var trainWithRelations = new ItemRatingWithRelationReader(trainReader, relations);
var testWithRelations = new ItemRatingWithRelationReader(testReader, relations);
var dataset = new Dataset <ItemRatingWithRelations>(trainWithRelations, testWithRelations);
Console.WriteLine("Features constructed.");
// step 2: recommender
var recommender = new LibFmTrainTester();
// step3: evaluation
var context = new EvalutationContext <ItemRating>(recommender, dataset);
var ep = new EvaluationPipeline <ItemRating>(context);
ep.Evaluators.Add(new RMSE());
ep.Run();
}
public void TestMovieLensSingle()
{
// step 1: dataset
var config = new CsvConfiguration()
{
Delimiter = "::",
HasHeaderRecord = true
};
// load data
var trainReader = new CsvReader(Paths.MovieLens1MTrain75, config);
var testReader = new CsvReader(Paths.MovieLens1MTest25, config, true);
var container = new DataContainer();
trainReader.LoadData(container);
testReader.LoadData(container);
var startTime = DateTime.Now;
var splitter = new RatingSimpleSplitter(container);
//var recommender = new MediaLiteRatingPredictor(new MatrixFactorization());
var recommender = new LibFmTrainTester(libFmPath: "LibFm.Net.64.exe");
// evaluation
var ctx = new EvalutationContext <ItemRating>(recommender, splitter);
var ep = new EvaluationPipeline <ItemRating>(ctx);
ep.Evaluators.Add(new RMSE());
ep.Run();
var duration = (int)DateTime.Now.Subtract(startTime).TotalMilliseconds;
Console.WriteLine("RMSE\tDuration\n{0}\t{1}", ctx["RMSE"], duration);
}
public void TestMovieLensAllDomains(int numDomains)
{
var numAuxRatings = new List<int> { 1 };
var movieLensReader = new MovieLensCrossDomainReader(Paths.MovieLens1MMovies, Paths.MovieLens1M);
var container = new MovieLensCrossDomainContainer(numDomains, false);
movieLensReader.LoadData(container);
double[,] rmseMatrix = new double[numAuxRatings.Count, numDomains];
int[,] durationsMatrix = new int[numAuxRatings.Count, numDomains];
int[] numUsers = new int[numDomains];
int[] numItems = new int[numDomains];
int[] numRatings = new int[numDomains];
int domainIndex = 0;
foreach (Domain d in container.Domains.Values)
{
var targetDomain = container.SpecifyTargetDomain(d.Id);
Console.WriteLine("Target domain: {0}", d.ToString());
var splitter = new CrossDomainSimpleSplitter(container, 0.25f);
int numAuxIndex = 0;
foreach (var num in numAuxRatings)
{
var startTime = DateTime.Now;
LibFmTrainTester recommender;
CrossDomainLibFmFeatureBuilder featureBuilder = null;
if (num == 0)
{
recommender = new LibFmTrainTester(experimentId: num.ToString());
}
else
{
featureBuilder = new CrossDomainLibFmFeatureBuilder(targetDomain, num);
recommender = new LibFmTrainTester(experimentId: num.ToString(), featureBuilder: featureBuilder);
}
var ctx = new EvalutationContext<ItemRating>(recommender, splitter);
var ep = new EvaluationPipeline<ItemRating>(ctx);
ep.Evaluators.Add(new RMSE());
ep.Run();
var duration = DateTime.Now.Subtract(startTime);
rmseMatrix[numAuxIndex, domainIndex] = recommender.RMSE;
durationsMatrix[numAuxIndex, domainIndex] = (int)duration.TotalMilliseconds;
numAuxIndex++;
}
numUsers[domainIndex] = d.Ratings.Select(r => r.User.Id).Distinct().Count();
numItems[domainIndex] = d.Ratings.Select(r => r.Item.Id).Distinct().Count();
numRatings[domainIndex] = d.Ratings.Count;
domainIndex++;
}
// Write RMSEs
Console.WriteLine("\nRMSEs:\n");
string header = Enumerable.Range(1, numDomains).Select(i => "D" + i).Aggregate((a, b) => a + "\t" + b);
Console.WriteLine("Num aux. ratings\t" + header);
for (int i = 0; i < numAuxRatings.Count; i++)
{
Console.Write(numAuxRatings[i]);
for (int j = 0; j < numDomains; j++)
{
Console.Write("\t" + rmseMatrix[i, j]);
}
Console.WriteLine();
}
// Write domain statistics
string users = numUsers.Select(c => c.ToString()).Aggregate((a, b) => a + "\t" + b);
string items = numItems.Select(c => c.ToString()).Aggregate((a, b) => a + "\t" + b);
string ratings = numRatings.Select(c => c.ToString()).Aggregate((a, b) => a + "\t" + b);
Console.WriteLine();
Console.WriteLine("Num Users\t" + users);
Console.WriteLine("Num Items\t" + items);
Console.WriteLine("Num Ratings\t" + ratings);
// Write times
Console.WriteLine("\nTimes:\n");
header = Enumerable.Range(1, numDomains).Select(i => "T" + i).Aggregate((a, b) => a + "\t" + b);
Console.WriteLine("Num aux. ratings\t" + header);
for (int i = 0; i < numAuxRatings.Count; i++)
{
Console.Write(numAuxRatings[i]);
for (int j = 0; j < numDomains; j++)
{
Console.Write("\t" + durationsMatrix[i, j]);
}
Console.WriteLine();
}
Console.WriteLine("\n");
}
public void TestAmazonWithNmfCluster()
{
var trainReader = new AmazonReader(Paths.AmazonBooksTrain75, Paths.AmazonBooksUsersCluster + ".nmf.u", Paths.AmazonBooksUsersCluster + ".nmf.i");
var testReader = new AmazonReader(Paths.AmazonBooksTest25, Paths.AmazonBooksUsersCluster + ".nmf.u", Paths.AmazonBooksUsersCluster + ".nmf.i");
var dataset = new Dataset<ItemRatingWithClusters>(trainReader, testReader);
var recommender = new LibFmTrainTester();
var context = new EvalutationContext<ItemRating>(recommender, dataset);
var ep = new EvaluationPipeline<ItemRating>(context);
ep.Evaluators.Add(new RMSE());
ep.Evaluators.Add(new MAE());
ep.Run();
}
public void TestEpinionClusters()
{
List<string> rmses = new List<string>();
List<string> maes = new List<string>();
for (int i = 0; i < 15; i += 2)
{
string usersClusterFile = Paths.EpinionUsersCluster + i + ".csv";
string itemsClusterFile = Paths.EpinionItemsCluster + i + ".csv";
EpinionReader trainReader, testReader;
if (i == 0)
{
trainReader = new EpinionReader(Paths.EpinionTrain75);
testReader = new EpinionReader(Paths.EpinionTest25);
}
else
{
trainReader = new EpinionReader(Paths.EpinionTrain75, usersClusterFile, itemsClusterFile);
testReader = new EpinionReader(Paths.EpinionTest25, usersClusterFile, itemsClusterFile);
}
var dataset = new Dataset<EpinionItemRating>(trainReader, testReader);
var recommender = new LibFmTrainTester();
var context = new EvalutationContext<ItemRating>(recommender, dataset);
var ep = new EvaluationPipeline<ItemRating>(context);
ep.Evaluators.Add(new RMSE());
ep.Evaluators.Add(new MAE());
ep.Run();
rmses.Add(context["RMSE"].ToString());
maes.Add(context["MAE"].ToString());
}
Console.WriteLine("RMSEs--------------");
rmses.ForEach(Console.WriteLine);
Console.WriteLine("MAEs-------------");
maes.ForEach(Console.WriteLine);
}
public void TestMovieLensWithClusters()
{
// Todo: instead of loading the dataset every time add a updateCluster method to the Dataset<ItemRatingWithCluster>
List<string> rmses = new List<string>();
List<string> maes = new List<string>();
for (int i = 0; i < 15; i += 2)
{
string usersClusterFile = Paths.MovieLens1MUsersCluster + i + ".csv";
string itemsClusterFile = Paths.MovieLens1MItemsCluster + i + ".csv";
MovieLensReader trainReader, testReader;
if (i == 0)
{
trainReader = new MovieLensReader(Paths.MovieLens1MTrain75);
testReader = new MovieLensReader(Paths.MovieLens1MTest25);
}
else
{
trainReader = new MovieLensReader(Paths.MovieLens1MTrain75, usersClusterFile, itemsClusterFile);
testReader = new MovieLensReader(Paths.MovieLens1MTest25, usersClusterFile, itemsClusterFile);
}
var dataset = new Dataset<MovieLensItemRating>(trainReader, testReader);
var recommender = new LibFmTrainTester();
var context = new EvalutationContext<ItemRating>(recommender, dataset);
var ep = new EvaluationPipeline<ItemRating>(context);
ep.Evaluators.Add(new RMSE());
ep.Evaluators.Add(new MAE());
ep.Run();
rmses.Add(context["RMSE"].ToString());
maes.Add(context["MAE"].ToString());
}
Console.WriteLine("RMSEs--------------");
rmses.ForEach(Console.WriteLine);
Console.WriteLine("MAEs-------------");
maes.ForEach(Console.WriteLine);
}
public void TestNewDataset()
{
// step 1: dataset
var config = new CsvConfiguration()
{
Delimiter = ",",
HasHeaderRecord = true
};
var container = new CrossDomainDataContainer();
var bookDomain = new Domain("book", true);
var musicDomain = new Domain("music");
var dvdDomain = new Domain("dvd");
var videoDomain = new Domain("video");
var bookReader = new CsvReader("books_selected4.csv", config, bookDomain);
//var trainReader = new CsvReader("books_selected1_train.csv", config, bookDomain);
//var testReader = new CsvReader("books_selected1_test.csv", config, bookDomain, true);
var musicReader = new CsvReader(Paths.AmazonAllMusicRatings, config, musicDomain);
var dvdReader = new CsvReader(Paths.AmazonAllDvdRatings, config, dvdDomain);
var videoReader = new CsvReader(Paths.AmazonAllVideoRatings, config, videoDomain);
bookReader.LoadData(container);
//trainReader.LoadData(container);
//testReader.LoadData(container);
musicReader.LoadData(container);
//dvdReader.LoadData(container);
//videoReader.LoadData(container);
container.PrintStatistics();
//musicDomain.CacheUserData();
var splitter = new CrossDomainSimpleSplitter(container, 0.25f);
//splitter.SaveSplitsAsCsv("books_selected1_train.csv", "books_selected1_test.csv");
//return;
//var splitter = new RatingSimpleSplitter(container, 0.25f);
var numAuxRatings = new List<int> { 0, 1, 2, 3, 5, 7, 10 };
var rmse = new List<string>();
var mae = new List<string>();
var durations = new List<string>();
foreach (var num in numAuxRatings)
{
var startTime = DateTime.Now;
// step 2: recommender
LibFmTrainTester recommender;
CrossDomainLibFmFeatureBuilder featureBuilder = null;
if (num == 0)
{
recommender = new LibFmTrainTester(experimentId: num.ToString());
}
else
{
featureBuilder = new CrossDomainLibFmFeatureBuilder(bookDomain, num);
recommender = new LibFmTrainTester(experimentId: num.ToString(), featureBuilder: featureBuilder);
}
// step3: evaluation
var ctx = new EvalutationContext<ItemRating>(recommender, splitter);
var ep = new EvaluationPipeline<ItemRating>(ctx);
ep.Evaluators.Add(new RMSE());
ep.Evaluators.Add(new MAE());
ep.Run();
//File.WriteAllLines("maps.txt", featureBuilder.Mapper.OriginalIDs.Zip(featureBuilder.Mapper.InternalIDs, (f, s) => f + "\t" + s));
rmse.Add(recommender.RMSE.ToString());
mae.Add(ctx["MAE"].ToString());
var duration = DateTime.Now.Subtract(startTime);
durations.Add(((int)duration.TotalMilliseconds).ToString());
}
Console.WriteLine("NumAuxRatings\tRMSE\tMAE\tDuration");
for (int i = 0; i < numAuxRatings.Count(); i++)
{
Console.WriteLine("{0}\t{1}\t{2}\t{3}", numAuxRatings[i], rmse[i], mae[i], durations[i]);
}
}
public void TestAuxDataSize()
{
// step 1: dataset
var config = new CsvConfiguration()
{
Delimiter = ",",
HasHeaderRecord = true
};
var container = new CrossDomainDataContainer();
var bookDomain = new Domain("book", true);
var musicDomain = new Domain("music");
var dvdDomain = new Domain("dvd");
var videoDomain = new Domain("video");
var bookReader = new CsvReader("books_selected4.csv", config, bookDomain);
//var trainReader = new CsvReader("books_selected1_train.csv", config, bookDomain);
//var testReader = new CsvReader("books_selected1_test.csv", config, bookDomain, true);
var musicReader = new CsvReader(Paths.AmazonAllMusicRatings, config, musicDomain);
var dvdReader = new CsvReader(Paths.AmazonAllDvdRatings, config, dvdDomain);
var videoReader = new CsvReader(Paths.AmazonAllVideoRatings, config, videoDomain);
bookReader.LoadData(container);
//trainReader.LoadData(container);
//testReader.LoadData(container);
musicReader.LoadData(container);
//dvdReader.LoadData(container);
//videoReader.LoadData(container);
container.PrintStatistics();
var splitter = new CrossDomainSimpleSplitter(container, 0.25f);
//splitter.SaveSplitsAsCsv("books_selected1_train.csv", "books_selected1_test.csv");
var rmse = new List<string>();
var mae = new List<string>();
var durations = new List<string>();
for (int i = 0; i < 10; i++)
{
var startTime = DateTime.Now;
musicDomain.ActivateData(0.1f);
// step 2: recommender
LibFmTrainTester recommender;
CrossDomainLibFmFeatureBuilder featureBuilder = null;
featureBuilder = new CrossDomainLibFmFeatureBuilder(bookDomain, 10);
recommender = new LibFmTrainTester(experimentId: i.ToString(), featureBuilder: featureBuilder);
// step3: evaluation
var ctx = new EvalutationContext<ItemRating>(recommender, splitter);
var ep = new EvaluationPipeline<ItemRating>(ctx);
ep.Evaluators.Add(new RMSE());
ep.Evaluators.Add(new MAE());
ep.Run();
rmse.Add(recommender.RMSE.ToString());
mae.Add(ctx["MAE"].ToString());
var duration = DateTime.Now.Subtract(startTime);
durations.Add(((int)duration.TotalMilliseconds).ToString());
}
Console.WriteLine("NumAuxRatings\tRMSE\tMAE\tDuration");
for (int i = 0; i < 10; i++)
{
Console.WriteLine("{0}\t{1}\t{2}\t{3}", i, rmse[i], mae[i], durations[i]);
}
}
public void TestAmazonCrossDomainVideo()
{
// step 1: dataset
var config = new CsvConfiguration()
{
Delimiter = ",",
HasHeaderRecord = true
};
var container = new CrossDomainDataContainer();
var bookDomain = new Domain("book");
var musicDomain = new Domain("music");
var dvdDomain = new Domain("dvd");
var videoDomain = new Domain("video", true);
var trainReader = new CsvReader(Paths.AmazonVideoTrain75, config, videoDomain);
var testReader = new CsvReader(Paths.AmazonVideoTest25, config, videoDomain, true);
var musicReader = new CsvReader(Paths.AmazonMusicRatings, config, musicDomain);
var dvdReader = new CsvReader(Paths.AmazonDvdRatings, config, dvdDomain);
var bookReader = new CsvReader(Paths.AmazonBooksRatings, config, bookDomain);
//var tempReader = new LibFmReader(_ecirTrain, _ecirTest) { MainDomain = bookDomain, AuxDomain = musicDomain, UserDataPath = _musicUsersPath };
trainReader.LoadData(container);
testReader.LoadData(container);
musicReader.LoadData(container);
dvdReader.LoadData(container);
bookReader.LoadData(container);
//tempReader.LoadData(container);
//container.ShuffleDomains();
container.PrintStatistics();
//musicDomain.CacheUserData();
var splitter = new CrossDomainSimpleSplitter(container);
//var splitter = new RatingSimpleSplitter(container);
var numAuxRatings = new int[4] { 0, 1, 2, 3 };
var rmse = new List<string>();
var mae = new List<string>();
var durations = new List<string>();
foreach (var num in numAuxRatings)
{
var startTime = DateTime.Now;
// step 2: recommender
ITrainTester<ItemRating> recommender;
CrossDomainLibFmFeatureBuilder featureBuilder = null;
if (num == 0)
{
recommender = new LibFmTrainTester(experimentId: num.ToString());
}
else
{
featureBuilder = new CrossDomainLibFmFeatureBuilder(videoDomain, num);
//featureBuilder.LoadCachedUserData(_musicUsersPath);
recommender = new LibFmTrainTester(experimentId: num.ToString(), featureBuilder: featureBuilder);
}
// step3: evaluation
var ctx = new EvalutationContext<ItemRating>(recommender, splitter);
var ep = new EvaluationPipeline<ItemRating>(ctx);
ep.Evaluators.Add(new RMSE());
ep.Evaluators.Add(new MAE());
ep.Run();
//File.WriteAllLines("maps.txt", featureBuilder.Mapper.OriginalIDs.Zip(featureBuilder.Mapper.InternalIDs, (f, s) => f + "\t" + s));
rmse.Add(ctx["RMSE"].ToString());
mae.Add(ctx["MAE"].ToString());
var duration = DateTime.Now.Subtract(startTime);
durations.Add(((int)duration.TotalMilliseconds).ToString());
}
Console.WriteLine("NumAuxRatings\tRMSE\tMAE\tDuration");
for (int i = 0; i < numAuxRatings.Count(); i++)
{
Console.WriteLine("{0}\t{1}\t{2}\t{3}", numAuxRatings[i], rmse[i], mae[i], durations[i]);
}
}
public void TestAmazonAllDomains2()
{
// step 1: dataset
var config = new CsvConfiguration()
{
Delimiter = ",",
HasHeaderRecord = true
};
var container = new DataContainer();
var bookReader = new CsvReader(Paths.AmazonBooksRatings, config);
var musicReader = new CsvReader(Paths.AmazonMusicRatings, config);
var dvdReader = new CsvReader(Paths.AmazonDvdRatings, config);
var videoReader = new CsvReader(Paths.AmazonVideoRatings, config);
bookReader.LoadData(container);
musicReader.LoadData(container);
dvdReader.LoadData(container);
videoReader.LoadData(container);
var splitter = new RatingSimpleSplitter(container, 0.25f);
var startTime = DateTime.Now;
var recommender = new LibFmTrainTester();
//var recommender = new MediaLiteRatingPredictor(new MatrixFactorization());
// step3: evaluation
var ctx = new EvalutationContext<ItemRating>(recommender, splitter);
var ep = new EvaluationPipeline<ItemRating>(ctx);
ep.Evaluators.Add(new RMSE());
ep.Evaluators.Add(new MAE());
ep.Run();
var duration = DateTime.Now.Subtract(startTime);
Console.WriteLine("RMSE\t{0}\nDuration\t{1}", ctx["RMSE"], (int)duration.TotalMilliseconds);
}
public void TestAmazonCrossDomainMusic()
{
// step 1: dataset
var config = new CsvConfiguration()
{
Delimiter = ",",
HasHeaderRecord = true
};
var container = new CrossDomainDataContainer();
var bookDomain = new Domain("book");
var musicDomain = new Domain("music", true);
var dvdDomain = new Domain("dvd");
var videoDomain = new Domain("video");
var trainReader = new CsvReader(Paths.AmazonMusicTrain75, config, musicDomain);
var testReader = new CsvReader(Paths.AmazonMusicTest25, config, musicDomain, true);
var bookReader = new CsvReader(Paths.AmazonBooksRatings, config, bookDomain);
var dvdReader = new CsvReader(Paths.AmazonDvdRatings, config, dvdDomain);
var videoReader = new CsvReader(Paths.AmazonVideoRatings, config, videoDomain);
//var tempReader = new LibFmReader(_ecirTrain, _ecirTest) { MainDomain = bookDomain, AuxDomain = musicDomain, UserDataPath = _musicUsersPath };
trainReader.LoadData(container);
testReader.LoadData(container);
bookReader.LoadData(container);
dvdReader.LoadData(container);
videoReader.LoadData(container);
//tempReader.LoadData(container);
//container.ShuffleDomains();
container.PrintStatistics();
//musicDomain.CacheUserData();
var splitter = new CrossDomainSimpleSplitter(container);
//var splitter = new RatingSimpleSplitter(container);
var numAuxRatings = new List <int> {
0, 1, 2, 3
};
var rmse = new List <string>();
var mae = new List <string>();
var durations = new List <string>();
foreach (var num in numAuxRatings)
{
var startTime = DateTime.Now;
// step 2: recommender
ITrainTester <ItemRating> recommender;
CrossDomainLibFmFeatureBuilder featureBuilder = null;
if (num == 0)
{
recommender = new LibFmTrainTester(experimentId: num.ToString());
}
else
{
featureBuilder = new CrossDomainLibFmFeatureBuilder(musicDomain, num);
//featureBuilder.LoadCachedUserData(_musicUsersPath);
recommender = new LibFmTrainTester(experimentId: num.ToString(), featureBuilder: featureBuilder);
}
// step3: evaluation
var ctx = new EvalutationContext <ItemRating>(recommender, splitter);
var ep = new EvaluationPipeline <ItemRating>(ctx);
ep.Evaluators.Add(new RMSE());
ep.Evaluators.Add(new MAE());
ep.Run();
//File.WriteAllLines("maps.txt", featureBuilder.Mapper.OriginalIDs.Zip(featureBuilder.Mapper.InternalIDs, (f, s) => f + "\t" + s));
rmse.Add(ctx["RMSE"].ToString());
mae.Add(ctx["MAE"].ToString());
var duration = DateTime.Now.Subtract(startTime);
durations.Add(((int)duration.TotalMilliseconds).ToString());
}
Console.WriteLine("NumAuxRatings\tRMSE\tMAE\tDuration");
for (int i = 0; i < numAuxRatings.Count(); i++)
{
Console.WriteLine("{0}\t{1}\t{2}\t{3}", numAuxRatings[i], rmse[i], mae[i], durations[i]);
}
}
public void TestNewDataset()
{
// step 1: dataset
var config = new CsvConfiguration()
{
Delimiter = ",",
HasHeaderRecord = true
};
var container = new CrossDomainDataContainer();
var bookDomain = new Domain("book", true);
var musicDomain = new Domain("music");
var dvdDomain = new Domain("dvd");
var videoDomain = new Domain("video");
var bookReader = new CsvReader("books_selected4.csv", config, bookDomain);
//var trainReader = new CsvReader("books_selected1_train.csv", config, bookDomain);
//var testReader = new CsvReader("books_selected1_test.csv", config, bookDomain, true);
var musicReader = new CsvReader(Paths.AmazonAllMusicRatings, config, musicDomain);
var dvdReader = new CsvReader(Paths.AmazonAllDvdRatings, config, dvdDomain);
var videoReader = new CsvReader(Paths.AmazonAllVideoRatings, config, videoDomain);
bookReader.LoadData(container);
//trainReader.LoadData(container);
//testReader.LoadData(container);
musicReader.LoadData(container);
//dvdReader.LoadData(container);
//videoReader.LoadData(container);
container.PrintStatistics();
//musicDomain.CacheUserData();
var splitter = new CrossDomainSimpleSplitter(container, 0.25f);
//splitter.SaveSplitsAsCsv("books_selected1_train.csv", "books_selected1_test.csv");
//return;
//var splitter = new RatingSimpleSplitter(container, 0.25f);
var numAuxRatings = new List <int> {
0, 1, 2, 3, 5, 7, 10
};
var rmse = new List <string>();
var mae = new List <string>();
var durations = new List <string>();
foreach (var num in numAuxRatings)
{
var startTime = DateTime.Now;
// step 2: recommender
LibFmTrainTester recommender;
CrossDomainLibFmFeatureBuilder featureBuilder = null;
if (num == 0)
{
recommender = new LibFmTrainTester(experimentId: num.ToString());
}
else
{
featureBuilder = new CrossDomainLibFmFeatureBuilder(bookDomain, num);
recommender = new LibFmTrainTester(experimentId: num.ToString(), featureBuilder: featureBuilder);
}
// step3: evaluation
var ctx = new EvalutationContext <ItemRating>(recommender, splitter);
var ep = new EvaluationPipeline <ItemRating>(ctx);
ep.Evaluators.Add(new RMSE());
ep.Evaluators.Add(new MAE());
ep.Run();
//File.WriteAllLines("maps.txt", featureBuilder.Mapper.OriginalIDs.Zip(featureBuilder.Mapper.InternalIDs, (f, s) => f + "\t" + s));
rmse.Add(recommender.RMSE.ToString());
mae.Add(ctx["MAE"].ToString());
var duration = DateTime.Now.Subtract(startTime);
durations.Add(((int)duration.TotalMilliseconds).ToString());
}
Console.WriteLine("NumAuxRatings\tRMSE\tMAE\tDuration");
for (int i = 0; i < numAuxRatings.Count(); i++)
{
Console.WriteLine("{0}\t{1}\t{2}\t{3}", numAuxRatings[i], rmse[i], mae[i], durations[i]);
}
}
public void TestMovieLensSingleDomain()
{
int numDomains = 1;
// load data
var movieLensReader = new MovieLensCrossDomainReader(Paths.MovieLens1MMovies, Paths.MovieLens1M);
var container = new MovieLensCrossDomainContainer(numDomains);
movieLensReader.LoadData(container);
// set taget and active domains
var targetDomain = container.SpecifyTargetDomain("ml0");
container.PrintStatistics();
var startTime = DateTime.Now;
var splitter = new CrossDomainSimpleSplitter(container, 0.25f);
// recommender with non-CrossDomain feature builder
var model = new MatrixFactorization();
model.NumIter = 50;
model.NumFactors = 8;
model.Regularization = 0.1f;
//var recommender = new MediaLiteRatingPredictor(model);
var recommender = new LibFmTrainTester();
// evaluation
var ctx = new EvalutationContext<ItemRating>(recommender, splitter);
var ep = new EvaluationPipeline<ItemRating>(ctx);
ep.Evaluators.Add(new RMSE());
ep.Evaluators.Add(new MAE());
ep.Run();
var duration = (int)DateTime.Now.Subtract(startTime).TotalMilliseconds;
Console.WriteLine("RMSE\tDuration\n{0}\t{1}", ctx["RMSE"], duration);
//container.CreateClusterFiles(Paths.MovieLens1M.GetDirectoryPath() + "\\movies.clust.raw", Paths.MovieLens1M.GetDirectoryPath() + "\\movies.clust.feat");
//container.WriteClusters(Paths.MovieLens1M.GetDirectoryPath() + "\\movies.clust");
}
public void TestAuxDataSize()
{
// step 1: dataset
var config = new CsvConfiguration()
{
Delimiter = ",",
HasHeaderRecord = true
};
var container = new CrossDomainDataContainer();
var bookDomain = new Domain("book", true);
var musicDomain = new Domain("music");
var dvdDomain = new Domain("dvd");
var videoDomain = new Domain("video");
var bookReader = new CsvReader("books_selected4.csv", config, bookDomain);
//var trainReader = new CsvReader("books_selected1_train.csv", config, bookDomain);
//var testReader = new CsvReader("books_selected1_test.csv", config, bookDomain, true);
var musicReader = new CsvReader(Paths.AmazonAllMusicRatings, config, musicDomain);
var dvdReader = new CsvReader(Paths.AmazonAllDvdRatings, config, dvdDomain);
var videoReader = new CsvReader(Paths.AmazonAllVideoRatings, config, videoDomain);
bookReader.LoadData(container);
//trainReader.LoadData(container);
//testReader.LoadData(container);
musicReader.LoadData(container);
//dvdReader.LoadData(container);
//videoReader.LoadData(container);
container.PrintStatistics();
var splitter = new CrossDomainSimpleSplitter(container, 0.25f);
//splitter.SaveSplitsAsCsv("books_selected1_train.csv", "books_selected1_test.csv");
var rmse = new List <string>();
var mae = new List <string>();
var durations = new List <string>();
for (int i = 0; i < 10; i++)
{
var startTime = DateTime.Now;
musicDomain.ActivateData(0.1f);
// step 2: recommender
LibFmTrainTester recommender;
CrossDomainLibFmFeatureBuilder featureBuilder = null;
featureBuilder = new CrossDomainLibFmFeatureBuilder(bookDomain, 10);
recommender = new LibFmTrainTester(experimentId: i.ToString(), featureBuilder: featureBuilder);
// step3: evaluation
var ctx = new EvalutationContext <ItemRating>(recommender, splitter);
var ep = new EvaluationPipeline <ItemRating>(ctx);
ep.Evaluators.Add(new RMSE());
ep.Evaluators.Add(new MAE());
ep.Run();
rmse.Add(recommender.RMSE.ToString());
mae.Add(ctx["MAE"].ToString());
var duration = DateTime.Now.Subtract(startTime);
durations.Add(((int)duration.TotalMilliseconds).ToString());
}
Console.WriteLine("NumAuxRatings\tRMSE\tMAE\tDuration");
for (int i = 0; i < 10; i++)
{
Console.WriteLine("{0}\t{1}\t{2}\t{3}", i, rmse[i], mae[i], durations[i]);
}
}
public void TestEpinionAllDomains(int numDomains = 3)
{
var numAuxRatings = new List <int> {
0, 1, 2, 3, 4
};
var epinionsReader = new EpinionsCrossDomainReader(Paths.EpinionRoot + "Epinions RED");
//var domainPaths = Enumerable.Range(1, numDomains)
// .Select(i => string.Format("{0}Epinions RED\\Domains{1}-{2}.csv", Paths.EpinionRoot, numDomains, i)).ToArray();
//var epinionsReader = new EpinionsCrossDomainReader(domainPaths);
var container = new EpinionsCrossDomainDataContainer(numDomains);
epinionsReader.LoadData(container);
container.Domains.Remove("ep0");
double[,] rmseMatrix = new double[numAuxRatings.Count, numDomains];
int[,] durationsMatrix = new int[numAuxRatings.Count, numDomains];
int[] numUsers = new int[numDomains];
int[] numItems = new int[numDomains];
int[] numRatings = new int[numDomains];
int domainIndex = 0;
foreach (Domain d in container.Domains.Values)
{
var targetDomain = container.SpecifyTargetDomain(d.Id);
Console.WriteLine("Target domain: {0}", d.ToString());
var splitter = new CrossDomainSimpleSplitter(container, 0.25f);
int numAuxIndex = 0;
foreach (var num in numAuxRatings)
{
var startTime = DateTime.Now;
LibFmTrainTester recommender;
CrossDomainLibFmFeatureBuilder featureBuilder = null;
if (num == 0)
{
recommender = new LibFmTrainTester(experimentId: num.ToString());
}
else
{
featureBuilder = new CrossDomainLibFmFeatureBuilder(targetDomain, num);
recommender = new LibFmTrainTester(experimentId: num.ToString(), featureBuilder: featureBuilder);
}
var ctx = new EvalutationContext <ItemRating>(recommender, splitter);
var ep = new EvaluationPipeline <ItemRating>(ctx);
ep.Evaluators.Add(new RMSE());
ep.Run();
var duration = DateTime.Now.Subtract(startTime);
rmseMatrix[numAuxIndex, domainIndex] = recommender.RMSE;
durationsMatrix[numAuxIndex, domainIndex] = (int)duration.TotalMilliseconds;
numAuxIndex++;
}
numUsers[domainIndex] = d.Ratings.Select(r => r.User.Id).Distinct().Count();
numItems[domainIndex] = d.Ratings.Select(r => r.Item.Id).Distinct().Count();
numRatings[domainIndex] = d.Ratings.Count;
domainIndex++;
}
// Write RMSEs
Console.WriteLine("\nRMSEs:\n");
string header = Enumerable.Range(1, numDomains).Select(i => "D" + i).Aggregate((a, b) => a + "\t" + b);
Console.WriteLine("Num aux. ratings\t" + header);
for (int i = 0; i < numAuxRatings.Count; i++)
{
Console.Write(numAuxRatings[i]);
for (int j = 0; j < numDomains; j++)
{
Console.Write("\t" + rmseMatrix[i, j]);
}
Console.WriteLine();
}
// Write domain statistics
string users = numUsers.Select(c => c.ToString()).Aggregate((a, b) => a + "\t" + b);
string items = numItems.Select(c => c.ToString()).Aggregate((a, b) => a + "\t" + b);
string ratings = numRatings.Select(c => c.ToString()).Aggregate((a, b) => a + "\t" + b);
Console.WriteLine();
Console.WriteLine("Num Users\t" + users);
Console.WriteLine("Num Items\t" + items);
Console.WriteLine("Num Ratings\t" + ratings);
// Write times
Console.WriteLine("\nTimes:\n");
header = Enumerable.Range(1, numDomains).Select(i => "T" + i).Aggregate((a, b) => a + "\t" + b);
Console.WriteLine("Num aux. ratings\t" + header);
for (int i = 0; i < numAuxRatings.Count; i++)
{
Console.Write(numAuxRatings[i]);
for (int j = 0; j < numDomains; j++)
{
Console.Write("\t" + durationsMatrix[i, j]);
}
Console.WriteLine();
}
Console.WriteLine("\n");
}
