public void TestSaveAndLoadTreeFeaturizer()
{
int dataPointCount = 200;
var data = SamplesUtils.DatasetUtils.GenerateFloatLabelFloatFeatureVectorSamples(dataPointCount).ToList();
var dataView = ML.Data.LoadFromEnumerable(data);
dataView = ML.Data.Cache(dataView);
var trainerOptions = new FastForestRegressionTrainer.Options
{
NumberOfThreads = 1,
NumberOfTrees = 10,
NumberOfLeaves = 4,
MinimumExampleCountPerLeaf = 10,
FeatureColumnName = "Features",
LabelColumnName = "Label"
};
var options = new FastForestRegressionFeaturizationEstimator.Options()
{
InputColumnName = "Features",
TreesColumnName = "Trees",
LeavesColumnName = "Leaves",
PathsColumnName = "Paths",
TrainerOptions = trainerOptions
};
var pipeline = ML.Transforms.FeaturizeByFastForestRegression(options)
.Append(ML.Transforms.Concatenate("CombinedFeatures", "Features", "Trees", "Leaves", "Paths"))
.Append(ML.Regression.Trainers.Sdca("Label", "CombinedFeatures"));
var model = pipeline.Fit(dataView);
var prediction = model.Transform(dataView);
var metrics = ML.Regression.Evaluate(prediction);
Assert.True(metrics.MeanAbsoluteError < 0.25);
Assert.True(metrics.MeanSquaredError < 0.1);
// Save the trained model into file.
ITransformer loadedModel = null;
var tempPath = Path.GetTempFileName();
using (var file = new SimpleFileHandle(Env, tempPath, true, true))
{
using (var fs = file.CreateWriteStream())
ML.Model.Save(model, null, fs);
using (var fs = file.OpenReadStream())
loadedModel = ML.Model.Load(fs, out var schema);
}
var loadedPrediction = loadedModel.Transform(dataView);
var loadedMetrics = ML.Regression.Evaluate(loadedPrediction);
Assert.Equal(metrics.MeanAbsoluteError, loadedMetrics.MeanAbsoluteError);
Assert.Equal(metrics.MeanSquaredError, loadedMetrics.MeanSquaredError);
}
public void TreeEnsembleFeaturizingPipelineMulticlass()
{
int dataPointCount = 1000;
var data = SamplesUtils.DatasetUtils.GenerateRandomMulticlassClassificationExamples(dataPointCount).ToList();
var dataView = ML.Data.LoadFromEnumerable(data);
dataView = ML.Data.Cache(dataView);
var trainerOptions = new FastForestRegressionTrainer.Options
{
NumberOfThreads = 1,
NumberOfTrees = 10,
NumberOfLeaves = 4,
MinimumExampleCountPerLeaf = 10,
FeatureColumnName = "Features",
LabelColumnName = "FloatLabel",
ShuffleLabels = true
};
var options = new FastForestRegressionFeaturizationEstimator.Options()
{
InputColumnName = "Features",
TreesColumnName = "Trees",
LeavesColumnName = "Leaves",
PathsColumnName = "Paths",
TrainerOptions = trainerOptions
};
Action <RowWithKey, RowWithFloat> actionConvertKeyToFloat = (RowWithKey rowWithKey, RowWithFloat rowWithFloat) =>
{
rowWithFloat.FloatLabel = rowWithKey.KeyLabel == 0 ? float.NaN : rowWithKey.KeyLabel - 1;
};
var split = ML.Data.TrainTestSplit(dataView, 0.5);
var trainData = split.TrainSet;
var testData = split.TestSet;
var pipeline = ML.Transforms.Conversion.MapValueToKey("KeyLabel", "Label")
.Append(ML.Transforms.CustomMapping(actionConvertKeyToFloat, "KeyLabel"))
.Append(ML.Transforms.FeaturizeByFastForestRegression(options))
.Append(ML.Transforms.Concatenate("CombinedFeatures", "Trees", "Leaves", "Paths"))
.Append(ML.MulticlassClassification.Trainers.SdcaMaximumEntropy("KeyLabel", "CombinedFeatures"));
var model = pipeline.Fit(trainData);
var prediction = model.Transform(testData);
var metrics = ML.MulticlassClassification.Evaluate(prediction, labelColumnName: "KeyLabel");
Assert.True(metrics.MacroAccuracy > 0.6);
Assert.True(metrics.MicroAccuracy > 0.6);
}
public void TestFastForestRegressionFeaturizationInPipeline()
{
int dataPointCount = 200;
var data = SamplesUtils.DatasetUtils.GenerateFloatLabelFloatFeatureVectorSamples(dataPointCount).ToList();
var dataView = ML.Data.LoadFromEnumerable(data);
dataView = ML.Data.Cache(dataView);
var trainerOptions = new FastForestRegressionTrainer.Options
{
NumberOfThreads = 1,
NumberOfTrees = 10,
NumberOfLeaves = 4,
MinimumExampleCountPerLeaf = 10,
FeatureColumnName = "Features",
LabelColumnName = "Label"
};
var options = new FastForestRegressionFeaturizationEstimator.Options()
{
InputColumnName = "Features",
TreesColumnName = "Trees",
LeavesColumnName = "Leaves",
PathsColumnName = "Paths",
TrainerOptions = trainerOptions
};
var pipeline = ML.Transforms.FeaturizeByFastForestRegression(options)
.Append(ML.Transforms.Concatenate("CombinedFeatures", "Features", "Trees", "Leaves", "Paths"))
.Append(ML.Regression.Trainers.Sdca("Label", "CombinedFeatures"));
var model = pipeline.Fit(dataView);
var prediction = model.Transform(dataView);
var metrics = ML.Regression.Evaluate(prediction);
Assert.True(metrics.MeanAbsoluteError < 0.25);
Assert.True(metrics.MeanSquaredError < 0.1);
}
public void TestSaveAndLoadDoubleTreeFeaturizer()
{
int dataPointCount = 200;
var data = SamplesUtils.DatasetUtils.GenerateFloatLabelFloatFeatureVectorSamples(dataPointCount).ToList();
var dataView = ML.Data.LoadFromEnumerable(data);
dataView = ML.Data.Cache(dataView);
var trainerOptions = new FastForestRegressionTrainer.Options
{
NumberOfThreads = 1,
NumberOfTrees = 10,
NumberOfLeaves = 4,
MinimumExampleCountPerLeaf = 10,
FeatureColumnName = "Features",
LabelColumnName = "Label"
};
// Trains tree featurization on "Features" and applies on "CopiedFeatures".
var options = new FastForestRegressionFeaturizationEstimator.Options()
{
InputColumnName = "CopiedFeatures",
TrainerOptions = trainerOptions,
TreesColumnName = "OhMyTrees",
LeavesColumnName = "OhMyLeaves",
PathsColumnName = "OhMyPaths"
};
var pipeline = ML.Transforms.CopyColumns("CopiedFeatures", "Features")
.Append(ML.Transforms.FeaturizeByFastForestRegression(options))
.Append(ML.Transforms.Concatenate("CombinedFeatures", "Features", "OhMyTrees", "OhMyLeaves", "OhMyPaths"))
.Append(ML.Regression.Trainers.Sdca("Label", "CombinedFeatures"));
var model = pipeline.Fit(dataView);
var prediction = model.Transform(dataView);
var metrics = ML.Regression.Evaluate(prediction);
Assert.True(metrics.MeanAbsoluteError < 0.25);
Assert.True(metrics.MeanSquaredError < 0.1);
// Save the trained model into file and then load it back.
ITransformer loadedModel = null;
var tempPath = Path.GetTempFileName();
using (var file = new SimpleFileHandle(Env, tempPath, true, true))
{
using (var fs = file.CreateWriteStream())
ML.Model.Save(model, null, fs);
using (var fs = file.OpenReadStream())
loadedModel = ML.Model.Load(fs, out var schema);
}
// Compute prediction using the loaded model.
var loadedPrediction = loadedModel.Transform(dataView);
var loadedMetrics = ML.Regression.Evaluate(loadedPrediction);
// Check if the loaded model produces the same result as the trained model.
Assert.Equal(metrics.MeanAbsoluteError, loadedMetrics.MeanAbsoluteError);
Assert.Equal(metrics.MeanSquaredError, loadedMetrics.MeanSquaredError);
var secondPipeline = ML.Transforms.CopyColumns("CopiedFeatures", "Features")
.Append(ML.Transforms.NormalizeBinning("CopiedFeatures"))
.Append(ML.Transforms.FeaturizeByFastForestRegression(options))
.Append(ML.Transforms.Concatenate("CombinedFeatures", "Features", "OhMyTrees", "OhMyLeaves", "OhMyPaths"))
.Append(ML.Regression.Trainers.Sdca("Label", "CombinedFeatures"));
var secondModel = secondPipeline.Fit(dataView);
var secondPrediction = secondModel.Transform(dataView);
var secondMetrics = ML.Regression.Evaluate(secondPrediction);
// The second pipeline trains a tree featurizer on a bin-based normalized feature, so the second pipeline
// is different from the first pipeline.
Assert.NotEqual(metrics.MeanAbsoluteError, secondMetrics.MeanAbsoluteError);
Assert.NotEqual(metrics.MeanSquaredError, secondMetrics.MeanSquaredError);
}
/// <summary>
/// Create <see cref="FastForestRegressionFeaturizationEstimator"/>, which uses <see cref="FastForestRegressionTrainer"/> to train <see cref="TreeEnsembleModelParameters"/> to create tree-based features.
/// </summary>
/// <param name="catalog">The context <see cref="TransformsCatalog"/> to create <see cref="PretrainedTreeFeaturizationEstimator"/>.</param>
/// <param name="options">The options to configure <see cref="FastForestRegressionFeaturizationEstimator"/>. See <see cref="FastForestRegressionFeaturizationEstimator.Options"/> and
/// <see cref="TreeEnsembleFeaturizationEstimatorBase.OptionsBase"/> for available settings.</param>
/// <example>
/// <format type="text/markdown">
/// <]
/// ]]>
/// </format>
/// </example>
public static FastForestRegressionFeaturizationEstimator FeaturizeByFastForestRegression(this TransformsCatalog catalog,
FastForestRegressionFeaturizationEstimator.Options options)
{
Contracts.CheckValue(catalog, nameof(catalog));
var env = CatalogUtils.GetEnvironment(catalog);
return(new FastForestRegressionFeaturizationEstimator(env, options));
}
// This example requires installation of additional NuGet package
// <a href="https://www.nuget.org/packages/Microsoft.ML.FastTree/">Microsoft.ML.FastTree</a>.
public static void Example()
{
// Create a new context for ML.NET operations. It can be used for
// exception tracking and logging, as a catalog of available operations
// and as the source of randomness. Setting the seed to a fixed number
// in this example to make outputs deterministic.
var mlContext = new MLContext(seed: 0);
// Create a list of training data points.
var dataPoints = GenerateRandomDataPoints(100).ToList();
// Convert the list of data points to an IDataView object, which is
// consumable by ML.NET API.
var dataView = mlContext.Data.LoadFromEnumerable(dataPoints);
// ML.NET doesn't cache data set by default. Therefore, if one reads a
// data set from a file and accesses it many times, it can be slow due
// to expensive featurization and disk operations. When the considered
// data can fit into memory, a solution is to cache the data in memory.
// Caching is especially helpful when working with iterative algorithms
// which needs many data passes.
dataView = mlContext.Data.Cache(dataView);
// Define input and output columns of tree-based featurizer.
string labelColumnName = nameof(DataPoint.Label);
string featureColumnName = nameof(DataPoint.Features);
string treesColumnName = nameof(TransformedDataPoint.Trees);
string leavesColumnName = nameof(TransformedDataPoint.Leaves);
string pathsColumnName = nameof(TransformedDataPoint.Paths);
// Define the configuration of the trainer used to train a tree-based
// model.
var trainerOptions = new FastForestRegressionTrainer.Options
{
// Only use 80% of features to reduce over-fitting.
FeatureFraction = 0.8,
// Create a simpler model by penalizing usage of new features.
FeatureFirstUsePenalty = 0.1,
// Reduce the number of trees to 3.
NumberOfTrees = 3,
// Number of leaves per tree.
NumberOfLeaves = 6,
LabelColumnName = labelColumnName,
FeatureColumnName = featureColumnName
};
// Define the tree-based featurizer's configuration.
var options = new FastForestRegressionFeaturizationEstimator.Options
{
InputColumnName = featureColumnName,
TreesColumnName = treesColumnName,
LeavesColumnName = leavesColumnName,
PathsColumnName = pathsColumnName,
TrainerOptions = trainerOptions
};
// Define the featurizer.
var pipeline = mlContext.Transforms.FeaturizeByFastForestRegression(
options);
// Train the model.
var model = pipeline.Fit(dataView);
// Create testing data. Use different random seed to make it different
// from training data.
var transformed = model.Transform(dataView);
// Convert IDataView object to a list. Each element in the resulted list
// corresponds to a row in the IDataView.
var transformedDataPoints = mlContext.Data.CreateEnumerable <
TransformedDataPoint>(transformed, false).ToList();
// Print out the transformation of the first 3 data points.
for (int i = 0; i < 3; ++i)
{
var dataPoint = dataPoints[i];
var transformedDataPoint = transformedDataPoints[i];
Console.WriteLine("The original feature vector [" + String.Join(",",
dataPoint.Features) + "] is transformed to three different " +
"tree-based feature vectors:");
Console.WriteLine(" Trees' output values: [" + String.Join(",",
transformedDataPoint.Trees) + "].");
Console.WriteLine(" Leave IDs' 0-1 representation: [" + String
.Join(",", transformedDataPoint.Leaves) + "].");
Console.WriteLine(" Paths IDs' 0-1 representation: [" + String
.Join(",", transformedDataPoint.Paths) + "].");
}
// Expected output:
// The original feature vector[1.543569, 1.494266, 1.284405] is
// transformed to three different tree - based feature vectors:
// Trees' output values: [0.7291142,0.7825329,0.8764582].
// Leave IDs' 0-1 representation: [0,1,0,0,0,0,0,0,1,0,0,0,0,0,1,0,0].
// Paths IDs' 0-1 representation: [1,0,0,0,1,1,1,0,0,1,1,1,0,1].
// The original feature vector[0.764918, 1.11206, 0.648211] is
// transformed to three different tree - based feature vectors:
// Trees' output values: [0.3802337,0.584159,0.5648927].
// Leave IDs' 0-1 representation: [0,0,1,0,0,0,0,1,0,0,0,0,0,1,0,0,0].
// Paths IDs' 0-1 representation: [1,1,1,0,0,1,1,0,0,0,1,1,0,0].
// The original feature vector[1.251254, 1.269456, 1.444864] is
// transformed to three different tree - based feature vectors:
// Trees' output values: [0.7591804,0.7825329,0.7443035].
// Leave IDs' 0-1 representation: [0,0,0,0,0,1,0,0,1,0,0,0,0,0,0,0,1].
// Paths IDs' 0-1 representation: [1,0,0,0,1,1,1,0,0,1,1,1,0,1].
}
