public void AutoNormalizationAndCaching()
{
var dataPath = GetDataPath(SentimentDataPath);
var testDataPath = GetDataPath(SentimentTestPath);
using (var env = new LocalEnvironment(seed: 1, conc: 1))
{
// Pipeline.
var loader = TextLoader.ReadFile(env, MakeSentimentTextLoaderArgs(), new MultiFileSource(dataPath));
var trans = TextTransform.Create(env, MakeSentimentTextTransformArgs(false), loader);
// Train.
var trainer = new LinearClassificationTrainer(env, new LinearClassificationTrainer.Arguments
{
NumThreads = 1,
ConvergenceTolerance = 1f
});
// Auto-caching.
IDataView trainData = trainer.Info.WantCaching ? (IDataView) new CacheDataView(env, trans, prefetch: null) : trans;
var trainRoles = new RoleMappedData(trainData, label: "Label", feature: "Features");
// Auto-normalization.
NormalizeTransform.CreateIfNeeded(env, ref trainRoles, trainer);
var predictor = trainer.Train(new Runtime.TrainContext(trainRoles));
}
}
void DecomposableTrainAndPredict()
{
using (var env = new LocalEnvironment()
.AddStandardComponents()) // ScoreUtils.GetScorer requires scorers to be registered in the ComponentCatalog
{
var loader = TextLoader.ReadFile(env, MakeIrisTextLoaderArgs(), new MultiFileSource(GetDataPath(TestDatasets.irisData.trainFilename)));
var term = TermTransform.Create(env, loader, "Label");
var concat = new ConcatTransform(env, "Features", "SepalLength", "SepalWidth", "PetalLength", "PetalWidth").Transform(term);
var trainer = new SdcaMultiClassTrainer(env, "Features", "Label", advancedSettings: (s) => { s.MaxIterations = 100; s.Shuffle = true; s.NumThreads = 1; });
IDataView trainData = trainer.Info.WantCaching ? (IDataView) new CacheDataView(env, concat, prefetch: null) : concat;
var trainRoles = new RoleMappedData(trainData, label: "Label", feature: "Features");
// Auto-normalization.
NormalizeTransform.CreateIfNeeded(env, ref trainRoles, trainer);
var predictor = trainer.Train(new Runtime.TrainContext(trainRoles));
var scoreRoles = new RoleMappedData(concat, label: "Label", feature: "Features");
IDataScorerTransform scorer = ScoreUtils.GetScorer(predictor, scoreRoles, env, trainRoles.Schema);
// Cut out term transform from pipeline.
var newScorer = ApplyTransformUtils.ApplyAllTransformsToData(env, scorer, loader, term);
var keyToValue = new KeyToValueTransform(env, "PredictedLabel").Transform(newScorer);
var model = env.CreatePredictionEngine <IrisDataNoLabel, IrisPrediction>(keyToValue);
var testData = loader.AsEnumerable <IrisDataNoLabel>(env, false);
foreach (var input in testData.Take(20))
{
var prediction = model.Predict(input);
Assert.True(prediction.PredictedLabel == "Iris-setosa");
}
}
}
// Returns true if a normalizer was added.
public static bool AddNormalizerIfNeeded(IHostEnvironment env, IChannel ch, ITrainer trainer, ref IDataView view, string featureColumn, NormalizeOption autoNorm)
{
Contracts.CheckValue(env, nameof(env));
env.CheckValue(ch, nameof(ch));
ch.CheckValue(trainer, nameof(trainer));
ch.CheckValue(view, nameof(view));
ch.CheckValueOrNull(featureColumn);
ch.CheckUserArg(Enum.IsDefined(typeof(NormalizeOption), autoNorm), nameof(TrainCommand.Arguments.NormalizeFeatures),
"Normalize option is invalid. Specify one of 'norm=No', 'norm=Warn', 'norm=Auto', or 'norm=Yes'.");
if (autoNorm == NormalizeOption.No)
{
ch.Info("Not adding a normalizer.");
return(false);
}
if (string.IsNullOrEmpty(featureColumn))
{
return(false);
}
int featCol;
var schema = view.Schema;
if (schema.TryGetColumnIndex(featureColumn, out featCol))
{
if (autoNorm != NormalizeOption.Yes)
{
if (!trainer.Info.NeedNormalization || schema[featCol].IsNormalized())
{
ch.Info("Not adding a normalizer.");
return(false);
}
if (autoNorm == NormalizeOption.Warn)
{
ch.Warning("A normalizer is needed for this trainer. Either add a normalizing transform or use the 'norm=Auto', 'norm=Yes' or 'norm=No' options.");
return(false);
}
}
ch.Info("Automatically adding a MinMax normalization transform, use 'norm=Warn' or 'norm=No' to turn this behavior off.");
IDataView ApplyNormalizer(IHostEnvironment innerEnv, IDataView input)
=> NormalizeTransform.CreateMinMaxNormalizer(innerEnv, input, featureColumn);
if (view is IDataLoader loader)
{
view = CompositeDataLoader.ApplyTransform(env, loader, tag: null, creationArgs: null, ApplyNormalizer);
}
else
{
view = ApplyNormalizer(env, view);
}
return(true);
}
return(false);
}
public void TrainAndPredictIrisModelUsingDirectInstantiationTest()
{
string dataPath = GetDataPath("iris.txt");
string testDataPath = dataPath;
using (var env = new TlcEnvironment(seed: 1, conc: 1))
{
// Pipeline
var loader = new TextLoader(env,
new TextLoader.Arguments()
{
HasHeader = false,
Column = new[]
{
new TextLoader.Column("Label", DataKind.R4, 0),
new TextLoader.Column("SepalLength", DataKind.R4, 1),
new TextLoader.Column("SepalWidth", DataKind.R4, 2),
new TextLoader.Column("PetalLength", DataKind.R4, 3),
new TextLoader.Column("PetalWidth", DataKind.R4, 4)
}
}, new MultiFileSource(dataPath));
IDataTransform trans = new ConcatTransform(env, loader, "Features",
"SepalLength", "SepalWidth", "PetalLength", "PetalWidth");
// Normalizer is not automatically added though the trainer has 'NormalizeFeatures' On/Auto
trans = NormalizeTransform.CreateMinMaxNormalizer(env, trans, "Features");
// Train
var trainer = new SdcaMultiClassTrainer(env, new SdcaMultiClassTrainer.Arguments()
{
NumThreads = 1
});
// Explicity adding CacheDataView since caching is not working though trainer has 'Caching' On/Auto
var cached = new CacheDataView(env, trans, prefetch: null);
var trainRoles = new RoleMappedData(cached, label: "Label", feature: "Features");
var pred = trainer.Train(trainRoles);
// Get scorer and evaluate the predictions from test data
IDataScorerTransform testDataScorer = GetScorer(env, trans, pred, testDataPath);
var metrics = Evaluate(env, testDataScorer);
CompareMatrics(metrics);
// Create prediction engine and test predictions
var model = env.CreatePredictionEngine <IrisData, IrisPrediction>(testDataScorer);
ComparePredictions(model);
// Get feature importance i.e. weight vector
var summary = ((MulticlassLogisticRegressionPredictor)pred).GetSummaryInKeyValuePairs(trainRoles.Schema);
Assert.Equal(7.757864, Convert.ToDouble(summary[0].Value), 5);
}
}
void Extensibility()
{
var dataPath = GetDataPath(IrisDataPath);
using (var env = new LocalEnvironment())
{
var loader = TextLoader.ReadFile(env, MakeIrisTextLoaderArgs(), new MultiFileSource(dataPath));
Action <IrisData, IrisData> action = (i, j) =>
{
j.Label = i.Label;
j.PetalLength = i.SepalLength > 3 ? i.PetalLength : i.SepalLength;
j.PetalWidth = i.PetalWidth;
j.SepalLength = i.SepalLength;
j.SepalWidth = i.SepalWidth;
};
var lambda = LambdaTransform.CreateMap(env, loader, action);
var term = TermTransform.Create(env, lambda, "Label");
var concat = new ConcatTransform(env, "Features", "SepalLength", "SepalWidth", "PetalLength", "PetalWidth")
.Transform(term);
var trainer = new SdcaMultiClassTrainer(env, new SdcaMultiClassTrainer.Arguments {
MaxIterations = 100, Shuffle = true, NumThreads = 1
});
IDataView trainData = trainer.Info.WantCaching ? (IDataView) new CacheDataView(env, concat, prefetch: null) : concat;
var trainRoles = new RoleMappedData(trainData, label: "Label", feature: "Features");
// Auto-normalization.
NormalizeTransform.CreateIfNeeded(env, ref trainRoles, trainer);
var predictor = trainer.Train(new Runtime.TrainContext(trainRoles));
var scoreRoles = new RoleMappedData(concat, label: "Label", feature: "Features");
IDataScorerTransform scorer = ScoreUtils.GetScorer(predictor, scoreRoles, env, trainRoles.Schema);
var keyToValue = new KeyToValueTransform(env, "PredictedLabel").Transform(scorer);
var model = env.CreatePredictionEngine <IrisData, IrisPrediction>(keyToValue);
var testLoader = TextLoader.ReadFile(env, MakeIrisTextLoaderArgs(), new MultiFileSource(dataPath));
var testData = testLoader.AsEnumerable <IrisData>(env, false);
foreach (var input in testData.Take(20))
{
var prediction = model.Predict(input);
Assert.True(prediction.PredictedLabel == input.Label);
}
}
}
public void Metacomponents()
{
using (var env = new LocalEnvironment())
{
var loader = TextLoader.ReadFile(env, MakeIrisTextLoaderArgs(), new MultiFileSource(GetDataPath(TestDatasets.irisData.trainFilename)));
var term = TermTransform.Create(env, loader, "Label");
var concat = new ConcatTransform(env, "Features", "SepalLength", "SepalWidth", "PetalLength", "PetalWidth").Transform(term);
var trainer = new Ova(env, new Ova.Arguments
{
PredictorType = ComponentFactoryUtils.CreateFromFunction(
e => new AveragedPerceptronTrainer(env, new AveragedPerceptronTrainer.Arguments()))
});
IDataView trainData = trainer.Info.WantCaching ? (IDataView) new CacheDataView(env, concat, prefetch: null) : concat;
var trainRoles = new RoleMappedData(trainData, label: "Label", feature: "Features");
// Auto-normalization.
NormalizeTransform.CreateIfNeeded(env, ref trainRoles, trainer);
var predictor = trainer.Train(new TrainContext(trainRoles));
}
}
void Metacomponents()
{
var dataPath = GetDataPath(IrisDataPath);
using (var env = new TlcEnvironment())
{
var loader = new TextLoader(env, MakeIrisTextLoaderArgs(), new MultiFileSource(dataPath));
var term = new TermTransform(env, loader, "Label");
var concat = new ConcatTransform(env, term, "Features", "SepalLength", "SepalWidth", "PetalLength", "PetalWidth");
var trainer = new Ova(env, new Ova.Arguments
{
PredictorType = new SimpleComponentFactory <ITrainer <IPredictorProducing <float> > >
(
(e) => new FastTreeBinaryClassificationTrainer(e, new FastTreeBinaryClassificationTrainer.Arguments())
)
});
IDataView trainData = trainer.Info.WantCaching ? (IDataView) new CacheDataView(env, concat, prefetch: null) : concat;
var trainRoles = new RoleMappedData(trainData, label: "Label", feature: "Features");
// Auto-normalization.
NormalizeTransform.CreateIfNeeded(env, ref trainRoles, trainer);
var predictor = trainer.Train(new Runtime.TrainContext(trainRoles));
var scoreRoles = new RoleMappedData(concat, label: "Label", feature: "Features");
IDataScorerTransform scorer = ScoreUtils.GetScorer(predictor, scoreRoles, env, trainRoles.Schema);
var keyToValue = new KeyToValueTransform(env, scorer, "PredictedLabel");
var model = env.CreatePredictionEngine <IrisData, IrisPrediction>(keyToValue);
var testLoader = new TextLoader(env, MakeIrisTextLoaderArgs(), new MultiFileSource(dataPath));
var testData = testLoader.AsEnumerable <IrisData>(env, false);
foreach (var input in testData.Take(20))
{
var prediction = model.Predict(input);
Assert.True(prediction.PredictedLabel == input.Label);
}
}
}
public void Metacomponents()
{
var dataPath = GetDataPath(IrisDataPath);
using (var env = new TlcEnvironment())
{
var loader = new TextLoader(env, MakeIrisTextLoaderArgs(), new MultiFileSource(dataPath));
var term = new TermTransform(env, loader, "Label");
var concat = new ConcatTransform(env, term, "Features", "SepalLength", "SepalWidth", "PetalLength", "PetalWidth");
var trainer = new Ova(env, new Ova.Arguments
{
PredictorType = ComponentFactoryUtils.CreateFromFunction(
e => new FastTreeBinaryClassificationTrainer(e, new FastTreeBinaryClassificationTrainer.Arguments()))
});
IDataView trainData = trainer.Info.WantCaching ? (IDataView) new CacheDataView(env, concat, prefetch: null) : concat;
var trainRoles = new RoleMappedData(trainData, label: "Label", feature: "Features");
// Auto-normalization.
NormalizeTransform.CreateIfNeeded(env, ref trainRoles, trainer);
var predictor = trainer.Train(new TrainContext(trainRoles));
}
}
void DecomposableTrainAndPredict()
{
var dataPath = GetDataPath(IrisDataPath);
using (var env = new TlcEnvironment())
{
var loader = new TextLoader(env, MakeIrisTextLoaderArgs(), new MultiFileSource(dataPath));
var term = new TermTransform(env, loader, "Label");
var concat = new ConcatTransform(env, term, "Features", "SepalLength", "SepalWidth", "PetalLength", "PetalWidth");
var trainer = new SdcaMultiClassTrainer(env, new SdcaMultiClassTrainer.Arguments {
MaxIterations = 100, Shuffle = true, NumThreads = 1
});
IDataView trainData = trainer.Info.WantCaching ? (IDataView) new CacheDataView(env, concat, prefetch: null) : concat;
var trainRoles = new RoleMappedData(trainData, label: "Label", feature: "Features");
// Auto-normalization.
NormalizeTransform.CreateIfNeeded(env, ref trainRoles, trainer);
var predictor = trainer.Train(new Runtime.TrainContext(trainRoles));
var scoreRoles = new RoleMappedData(concat, label: "Label", feature: "Features");
IDataScorerTransform scorer = ScoreUtils.GetScorer(predictor, scoreRoles, env, trainRoles.Schema);
// Cut out term transform from pipeline.
var newScorer = ApplyTransformUtils.ApplyAllTransformsToData(env, scorer, loader, term);
var keyToValue = new KeyToValueTransform(env, newScorer, "PredictedLabel");
var model = env.CreatePredictionEngine <IrisDataNoLabel, IrisPrediction>(keyToValue);
var testLoader = new TextLoader(env, MakeIrisTextLoaderArgs(), new MultiFileSource(dataPath));
var testData = testLoader.AsEnumerable <IrisDataNoLabel>(env, false);
foreach (var input in testData.Take(20))
{
var prediction = model.Predict(input);
Assert.True(prediction.PredictedLabel == "Iris-setosa");
}
}
}
protected TTransformer TrainTransformer(IDataView trainSet,
IDataView validationSet = null, IPredictor initPredictor = null)
{
var cachedTrain = TrainerInfo.WantCaching ? new CacheDataView(_env, trainSet, prefetch: null) : trainSet;
var trainRoles = new RoleMappedData(cachedTrain, label: _labelCol, feature: _featureCol);
var emptyData = new EmptyDataView(_env, trainSet.Schema);
IDataView normalizer = emptyData;
if (TrainerInfo.NeedNormalization && trainRoles.Schema.FeaturesAreNormalized() == false)
{
var view = NormalizeTransform.CreateMinMaxNormalizer(_env, trainRoles.Data, name: trainRoles.Schema.Feature.Name);
normalizer = ApplyTransformUtils.ApplyAllTransformsToData(_env, view, emptyData, cachedTrain);
trainRoles = new RoleMappedData(view, trainRoles.Schema.GetColumnRoleNames());
}
RoleMappedData validRoles;
if (validationSet == null)
{
validRoles = null;
}
else
{
var cachedValid = TrainerInfo.WantCaching ? new CacheDataView(_env, validationSet, prefetch: null) : validationSet;
cachedValid = ApplyTransformUtils.ApplyAllTransformsToData(_env, normalizer, cachedValid);
validRoles = new RoleMappedData(cachedValid, label: _labelCol, feature: _featureCol);
}
var pred = TrainCore(new TrainContext(trainRoles, validRoles, initPredictor));
var scoreRoles = new RoleMappedData(normalizer, label: _labelCol, feature: _featureCol);
return(MakeScorer(pred, scoreRoles));
}
private static IPredictor TrainKMeansAndLRCore()
{
string dataPath = s_dataPath;
using (var env = new TlcEnvironment(seed: 1))
{
// Pipeline
var loader = new TextLoader(env,
new TextLoader.Arguments()
{
HasHeader = true,
Separator = ",",
Column = new[] {
new TextLoader.Column()
{
Name = "Label",
Source = new [] { new TextLoader.Range()
{
Min = 14, Max = 14
} },
Type = DataKind.R4
},
new TextLoader.Column()
{
Name = "CatFeatures",
Source = new [] {
new TextLoader.Range()
{
Min = 1, Max = 1
},
new TextLoader.Range()
{
Min = 3, Max = 3
},
new TextLoader.Range()
{
Min = 5, Max = 9
},
new TextLoader.Range()
{
Min = 13, Max = 13
}
},
Type = DataKind.TX
},
new TextLoader.Column()
{
Name = "NumFeatures",
Source = new [] {
new TextLoader.Range()
{
Min = 0, Max = 0
},
new TextLoader.Range()
{
Min = 2, Max = 2
},
new TextLoader.Range()
{
Min = 4, Max = 4
},
new TextLoader.Range()
{
Min = 10, Max = 12
}
},
Type = DataKind.R4
}
}
}, new MultiFileSource(dataPath));
IDataTransform trans = CategoricalTransform.Create(env, new CategoricalTransform.Arguments
{
Column = new[]
{
new CategoricalTransform.Column {
Name = "CatFeatures", Source = "CatFeatures"
}
}
}, loader);
trans = NormalizeTransform.CreateMinMaxNormalizer(env, trans, "NumFeatures");
trans = new ConcatTransform(env, trans, "Features", "NumFeatures", "CatFeatures");
trans = TrainAndScoreTransform.Create(env, new TrainAndScoreTransform.Arguments
{
Trainer = new SubComponent <ITrainer, SignatureTrainer>("KMeans", "k=100"),
FeatureColumn = "Features"
}, trans);
trans = new ConcatTransform(env, trans, "Features", "Features", "Score");
// Train
var trainer = new LogisticRegression(env, new LogisticRegression.Arguments()
{
EnforceNonNegativity = true, OptTol = 1e-3f
});
var trainRoles = new RoleMappedData(trans, label: "Label", feature: "Features");
return(trainer.Train(trainRoles));
}
}
public ParameterMixingCalibratedPredictor TrainKMeansAndLR()
{
using (var env = new ConsoleEnvironment(seed: 1))
{
// Pipeline
var loader = TextLoader.ReadFile(env,
new TextLoader.Arguments()
{
HasHeader = true,
Separator = ",",
Column = new[] {
new TextLoader.Column("Label", DataKind.R4, 14),
new TextLoader.Column("CatFeatures", DataKind.TX,
new [] {
new TextLoader.Range()
{
Min = 1, Max = 1
},
new TextLoader.Range()
{
Min = 3, Max = 3
},
new TextLoader.Range()
{
Min = 5, Max = 9
},
new TextLoader.Range()
{
Min = 13, Max = 13
}
}),
new TextLoader.Column("NumFeatures", DataKind.R4,
new [] {
new TextLoader.Range()
{
Min = 0, Max = 0
},
new TextLoader.Range()
{
Min = 2, Max = 2
},
new TextLoader.Range()
{
Min = 4, Max = 4
},
new TextLoader.Range()
{
Min = 10, Max = 12
}
})
}
}, new MultiFileSource(_dataPath));
IDataView trans = new CategoricalEstimator(env, "CatFeatures").Fit(loader).Transform(loader);
trans = NormalizeTransform.CreateMinMaxNormalizer(env, trans, "NumFeatures");
trans = new ConcatTransform(env, "Features", "NumFeatures", "CatFeatures").Transform(trans);
trans = TrainAndScoreTransform.Create(env, new TrainAndScoreTransform.Arguments
{
Trainer = ComponentFactoryUtils.CreateFromFunction(host =>
new KMeansPlusPlusTrainer(host, "Features", advancedSettings: s =>
{
s.K = 100;
})),
FeatureColumn = "Features"
}, trans);
trans = new ConcatTransform(env, "Features", "Features", "Score").Transform(trans);
// Train
var trainer = new LogisticRegression(env, "Features", "Label", advancedSettings: args => { args.EnforceNonNegativity = true; args.OptTol = 1e-3f; });
var trainRoles = new RoleMappedData(trans, label: "Label", feature: "Features");
return(trainer.Train(trainRoles));
}
}
void CrossValidation()
{
var dataPath = GetDataPath(SentimentDataPath);
var testDataPath = GetDataPath(SentimentTestPath);
int numFolds = 5;
using (var env = new TlcEnvironment(seed: 1, conc: 1))
{
// Pipeline.
var loader = new TextLoader(env, MakeSentimentTextLoaderArgs(), new MultiFileSource(dataPath));
var text = TextTransform.Create(env, MakeSentimentTextTransformArgs(false), loader);
IDataView trans = new GenerateNumberTransform(env, text, "StratificationColumn");
// Train.
var trainer = new LinearClassificationTrainer(env, new LinearClassificationTrainer.Arguments
{
NumThreads = 1,
ConvergenceTolerance = 1f
});
var metrics = new List <BinaryClassificationMetrics>();
for (int fold = 0; fold < numFolds; fold++)
{
IDataView trainPipe = new RangeFilter(env, new RangeFilter.Arguments()
{
Column = "StratificationColumn",
Min = (Double)fold / numFolds,
Max = (Double)(fold + 1) / numFolds,
Complement = true
}, trans);
trainPipe = new OpaqueDataView(trainPipe);
var trainData = new RoleMappedData(trainPipe, label: "Label", feature: "Features");
// Auto-normalization.
NormalizeTransform.CreateIfNeeded(env, ref trainData, trainer);
var preCachedData = trainData;
// Auto-caching.
if (trainer.Info.WantCaching)
{
var prefetch = trainData.Schema.GetColumnRoles().Select(kc => kc.Value.Index).ToArray();
var cacheView = new CacheDataView(env, trainData.Data, prefetch);
// Because the prefetching worked, we know that these are valid columns.
trainData = new RoleMappedData(cacheView, trainData.Schema.GetColumnRoleNames());
}
var predictor = trainer.Train(new Runtime.TrainContext(trainData));
IDataView testPipe = new RangeFilter(env, new RangeFilter.Arguments()
{
Column = "StratificationColumn",
Min = (Double)fold / numFolds,
Max = (Double)(fold + 1) / numFolds,
Complement = false
}, trans);
testPipe = new OpaqueDataView(testPipe);
var pipe = ApplyTransformUtils.ApplyAllTransformsToData(env, preCachedData.Data, testPipe, trainPipe);
var testRoles = new RoleMappedData(pipe, trainData.Schema.GetColumnRoleNames());
IDataScorerTransform scorer = ScoreUtils.GetScorer(predictor, testRoles, env, testRoles.Schema);
BinaryClassifierMamlEvaluator eval = new BinaryClassifierMamlEvaluator(env, new BinaryClassifierMamlEvaluator.Arguments()
{
});
var dataEval = new RoleMappedData(scorer, testRoles.Schema.GetColumnRoleNames(), opt: true);
var dict = eval.Evaluate(dataEval);
var foldMetrics = BinaryClassificationMetrics.FromMetrics(env, dict["OverallMetrics"], dict["ConfusionMatrix"]);
metrics.Add(foldMetrics.Single());
}
}
}
