machinelearning/test/Microsoft.ML.AutoML.Tests/AutoMLExperimentTests.cs

519 строки
21 KiB
C#

// Licensed to the .NET Foundation under one or more agreements.
// The .NET Foundation licenses this file to you under the MIT license.
// See the LICENSE file in the project root for more information.
using System;
using System.Collections.Generic;
using System.Diagnostics;
using System.IO;
using System.Linq;
using System.Text;
using System.Threading;
using System.Threading.Tasks;
using FluentAssertions;
using Microsoft.Data.Analysis;
using Microsoft.Extensions.DependencyInjection;
using Microsoft.ML.AutoML.CodeGen;
using Microsoft.ML.Fairlearn.AutoML;
using Microsoft.ML.Runtime;
using Microsoft.ML.TestFramework;
using Microsoft.ML.TestFramework.Attributes;
using Xunit;
using Xunit.Abstractions;
namespace Microsoft.ML.AutoML.Test
{
public class AutoMLExperimentTests : BaseTestClass
{
public AutoMLExperimentTests(ITestOutputHelper output) : base(output)
{
}
[Fact]
public async Task AutoMLExperiment_throw_timeout_exception_when_ct_is_canceled_and_no_trial_completed_Async()
{
var context = new MLContext(1);
var experiment = context.Auto().CreateExperiment();
experiment.SetTrainingTimeInSeconds(1)
.SetTrialRunner((serviceProvider) =>
{
var channel = serviceProvider.GetService<IChannel>();
var settings = serviceProvider.GetService<AutoMLExperiment.AutoMLExperimentSettings>();
return new DummyTrialRunner(settings, 5, channel);
})
.SetTuner<RandomSearchTuner>();
var cts = new CancellationTokenSource();
context.Log += (o, e) =>
{
if (e.RawMessage.Contains("Update Running Trial"))
{
cts.Cancel();
}
};
var runExperimentAction = async () => await experiment.RunAsync(cts.Token);
await runExperimentAction.Should().ThrowExactlyAsync<TimeoutException>();
}
[Fact]
public async Task AutoMLExperiment_cancel_trial_when_exceeds_memory_limit_Async()
{
var context = new MLContext(1);
var experiment = context.Auto().CreateExperiment();
context.Log += (o, e) =>
{
if (e.Source.StartsWith("AutoMLExperiment"))
{
this.Output.WriteLine(e.RawMessage);
}
};
// the following experiment set memory usage limit to 0.01mb
// so all trials should be canceled and there should be no successful trials.
// therefore when experiment finishes, it should throw timeout exception with no model trained message.
experiment.SetMaxModelToExplore(10)
.SetTrialRunner((serviceProvider) =>
{
var channel = serviceProvider.GetService<IChannel>();
var settings = serviceProvider.GetService<AutoMLExperiment.AutoMLExperimentSettings>();
return new DummyTrialRunner(settings, 5, channel);
})
.SetTuner<RandomSearchTuner>()
.SetMaximumMemoryUsageInMegaByte(0.01);
var runExperimentAction = async () => await experiment.RunAsync();
await runExperimentAction.Should().ThrowExactlyAsync<TimeoutException>();
}
[LightGBMFact]
public async Task AutoMLExperiment_lgbm_cancel_trial_when_exceeds_memory_limit_Async()
{
// this test is to verify that lightGbm can be cancelled during training booster.
var context = new MLContext(1);
context.Log += (o, e) =>
{
if (e.Message.Contains("LightGBM objective"))
{
context.CancelExecution();
}
};
var data = DatasetUtil.GetUciAdultDataView();
var experiment = context.Auto().CreateExperiment();
var pipeline = context.Auto().Featurizer(data, "_Features_", excludeColumns: new[] { DatasetUtil.UciAdultLabel })
.Append(context.BinaryClassification.Trainers.LightGbm(DatasetUtil.UciAdultLabel, "_Features_", numberOfIterations: 10000));
experiment.SetDataset(context.Data.TrainTestSplit(data))
.SetBinaryClassificationMetric(BinaryClassificationMetric.AreaUnderRocCurve, DatasetUtil.UciAdultLabel)
.SetPipeline(pipeline)
.SetTrainingTimeInSeconds(10)
.SetMaximumMemoryUsageInMegaByte(10);
var runExperimentAction = async () => await experiment.RunAsync();
await runExperimentAction.Should().ThrowExactlyAsync<TimeoutException>();
}
[Fact]
public async Task AutoMLExperiment_return_current_best_trial_when_ct_is_canceled_with_trial_completed_Async()
{
var context = new MLContext(1);
var stopWatch = new Stopwatch();
stopWatch.Start();
var experiment = context.Auto().CreateExperiment();
experiment.SetTrainingTimeInSeconds(10)
.SetTrialRunner((serviceProvider) =>
{
var channel = serviceProvider.GetService<IChannel>();
var settings = serviceProvider.GetService<AutoMLExperiment.AutoMLExperimentSettings>();
return new DummyTrialRunner(settings, 1, channel);
})
.SetTuner<RandomSearchTuner>();
var cts = new CancellationTokenSource();
context.Log += (o, e) =>
{
if (e.RawMessage.Contains("Update Completed Trial"))
{
cts.CancelAfter(100);
}
};
var res = await experiment.RunAsync(cts.Token);
stopWatch.Stop();
stopWatch.ElapsedMilliseconds.Should().BeLessOrEqualTo(2 * 1000 + 500);
cts.IsCancellationRequested.Should().BeTrue();
res.Metric.Should().BeGreaterThan(0);
}
[Fact]
public async Task AutoMLExperiment_finish_training_when_time_is_up_Async()
{
var context = new MLContext(1);
context.Log += (o, e) =>
{
if (e.Source.StartsWith("AutoMLExperiment"))
{
this.Output.WriteLine(e.RawMessage);
}
};
var experiment = context.Auto().CreateExperiment();
experiment.SetTrainingTimeInSeconds(5)
.SetTrialRunner((serviceProvider) =>
{
var channel = serviceProvider.GetService<IChannel>();
var settings = serviceProvider.GetService<AutoMLExperiment.AutoMLExperimentSettings>();
return new DummyTrialRunner(settings, 0, channel);
})
.SetTuner<RandomSearchTuner>();
var cts = new CancellationTokenSource();
cts.CancelAfter(10 * 1000);
var res = await experiment.RunAsync(cts.Token);
res.Metric.Should().BeGreaterThan(0);
cts.IsCancellationRequested.Should().BeFalse();
}
[Fact]
public async Task AutoMLExperiment_finish_training_when_reach_to_max_model_async()
{
var context = new MLContext(1);
var experiment = context.Auto().CreateExperiment();
experiment.SetMaxModelToExplore(5)
.SetTrialRunner((serviceProvider) =>
{
var channel = serviceProvider.GetService<IChannel>();
var settings = serviceProvider.GetService<AutoMLExperiment.AutoMLExperimentSettings>();
return new DummyTrialRunner(settings, 1, channel);
})
.SetTuner<RandomSearchTuner>();
var runModelCounts = 0;
context.Log += (o, e) =>
{
if (e.RawMessage.Contains("Update Completed Trial"))
{
runModelCounts++;
}
};
await experiment.RunAsync();
runModelCounts.Should().Be(5);
}
[Fact]
public async Task AutoMLExperiment_UCI_Adult_Train_Test_Split_Test()
{
var context = new MLContext(1);
context.Log += (o, e) =>
{
if (e.Source.StartsWith("AutoMLExperiment"))
{
this.Output.WriteLine(e.RawMessage);
}
};
var data = DatasetUtil.GetUciAdultDataView();
var experiment = context.Auto().CreateExperiment();
var pipeline = context.Auto().Featurizer(data, "_Features_", excludeColumns: new[] { DatasetUtil.UciAdultLabel })
.Append(context.Auto().BinaryClassification(DatasetUtil.UciAdultLabel, "_Features_", useLgbm: false, useSdcaLogisticRegression: false, useLbfgsLogisticRegression: false));
experiment.SetDataset(context.Data.TrainTestSplit(data))
.SetBinaryClassificationMetric(BinaryClassificationMetric.AreaUnderRocCurve, DatasetUtil.UciAdultLabel)
.SetPipeline(pipeline)
.SetMaxModelToExplore(1);
var result = await experiment.RunAsync();
result.Metric.Should().BeGreaterThan(0.8);
}
[Fact(Skip = "skip in CI build")]
public async Task AutoMLExperiment_UCI_Adult_Fairlearn_Test()
{
var context = new MLContext(1);
context.Log += (o, e) =>
{
if (e.Source.StartsWith("AutoMLExperiment"))
{
this.Output.WriteLine(e.RawMessage);
}
};
var data = DatasetUtil.GetUciAdultDataView();
var experiment = context.Auto().CreateExperiment();
var pipeline = context.Auto().Featurizer(data, "_Features_", excludeColumns: new[] { DatasetUtil.UciAdultLabel })
.Append(context.Auto().BinaryClassification(DatasetUtil.UciAdultLabel, "_Features_", exampleWeightColumnName: "signedWeight", useLgbm: false, useSdcaLogisticRegression: false, useLbfgsLogisticRegression: false));
experiment.SetDataset(context.Data.TrainTestSplit(data))
.SetPipeline(pipeline)
.SetBinaryClassificationMetricWithFairLearn(DatasetUtil.UciAdultLabel, "PredictedLabel", "Workclass", "signedWeight")
.SetMaxModelToExplore(100);
var result = await experiment.RunAsync();
result.Metric.Should().BeGreaterThan(0.8);
}
[Fact]
public async Task AutoMLExperiment_UCI_Adult_CV_5_Test()
{
var context = new MLContext(1);
context.Log += (o, e) =>
{
if (e.Source.StartsWith("AutoMLExperiment"))
{
this.Output.WriteLine(e.RawMessage);
}
};
var data = DatasetUtil.GetUciAdultDataView();
var experiment = context.Auto().CreateExperiment();
var pipeline = context.Auto().Featurizer(data, "_Features_", excludeColumns: new[] { DatasetUtil.UciAdultLabel })
.Append(context.Auto().BinaryClassification(DatasetUtil.UciAdultLabel, "_Features_", useLgbm: false, useSdcaLogisticRegression: false, useLbfgsLogisticRegression: false));
experiment.SetDataset(data, 5)
.SetBinaryClassificationMetric(BinaryClassificationMetric.AreaUnderRocCurve, DatasetUtil.UciAdultLabel)
.SetPipeline(pipeline)
.SetMaxModelToExplore(1);
var result = await experiment.RunAsync();
result.Metric.Should().BeGreaterThan(0.8);
}
[Fact]
public async Task AutoMLExperiment_Iris_CV_5_Test()
{
var context = new MLContext(1);
context.Log += (o, e) =>
{
if (e.RawMessage.Contains("Trial"))
{
this.Output.WriteLine(e.RawMessage);
}
};
var data = DatasetUtil.GetIrisDataView();
var experiment = context.Auto().CreateExperiment();
var label = "Label";
var pipeline = context.Auto().Featurizer(data, excludeColumns: new[] { label })
.Append(context.Transforms.Conversion.MapValueToKey(label, label))
.Append(context.Auto().MultiClassification(label, useLgbm: false, useSdcaMaximumEntrophy: false, useLbfgsMaximumEntrophy: false));
experiment.SetDataset(data, 5)
.SetMulticlassClassificationMetric(MulticlassClassificationMetric.MacroAccuracy, label)
.SetPipeline(pipeline)
.SetMaxModelToExplore(1);
var result = await experiment.RunAsync();
result.Metric.Should().BeGreaterThan(0.8);
}
[Fact]
public async Task AutoMLExperiment_Iris_Train_Test_Split_Test()
{
var context = new MLContext(1);
context.Log += (o, e) =>
{
if (e.Source.StartsWith("AutoMLExperiment"))
{
this.Output.WriteLine(e.RawMessage);
}
};
var data = DatasetUtil.GetIrisDataView();
var experiment = context.Auto().CreateExperiment();
var label = "Label";
var pipeline = context.Auto().Featurizer(data, excludeColumns: new[] { label })
.Append(context.Transforms.Conversion.MapValueToKey(label, label))
.Append(context.Auto().MultiClassification(label, useLgbm: false, useSdcaMaximumEntrophy: false, useLbfgsMaximumEntrophy: false));
experiment.SetDataset(context.Data.TrainTestSplit(data))
.SetMulticlassClassificationMetric(MulticlassClassificationMetric.MacroAccuracy, label)
.SetPipeline(pipeline)
.SetMaxModelToExplore(1);
var result = await experiment.RunAsync();
result.Metric.Should().BeGreaterThan(0.8);
}
[Fact]
public async Task AutoMLExperiment_Taxi_Fare_Train_Test_Split_Test()
{
var context = new MLContext(1);
context.Log += (o, e) =>
{
if (e.Source.StartsWith("AutoMLExperiment"))
{
this.Output.WriteLine(e.RawMessage);
}
};
var train = DatasetUtil.GetTaxiFareTrainDataView();
var test = DatasetUtil.GetTaxiFareTestDataView();
var experiment = context.Auto().CreateExperiment();
var label = DatasetUtil.TaxiFareLabel;
var pipeline = context.Auto().Featurizer(train, excludeColumns: new[] { label })
.Append(context.Auto().Regression(label, useLgbm: false, useSdca: false, useLbfgsPoissonRegression: false));
experiment.SetDataset(train, test)
.SetRegressionMetric(RegressionMetric.RSquared, label)
.SetPipeline(pipeline)
.SetMaxModelToExplore(1);
var result = await experiment.RunAsync();
result.Metric.Should().BeGreaterThan(0.5);
// test subsamping
experiment = context.Auto().CreateExperiment();
experiment.SetDataset(train, test, true)
.SetRegressionMetric(RegressionMetric.RSquared, label)
.SetPipeline(pipeline)
.SetMaxModelToExplore(1);
result = await experiment.RunAsync();
result.Metric.Should().BeGreaterThan(0.5);
result.TrialSettings.Parameter[nameof(TrainValidateDatasetManager)]["TrainValidateDatasetSubsamplingKey"]
.AsType<float>().Should().Be(0.1f);
}
[Fact]
public async Task AutoMLExperiment_Taxi_Fare_CV_5_Test()
{
var context = new MLContext(1);
var train = DatasetUtil.GetTaxiFareTrainDataView();
var experiment = context.Auto().CreateExperiment();
var label = DatasetUtil.TaxiFareLabel;
var pipeline = context.Auto().Featurizer(train, excludeColumns: new[] { label })
.Append(context.Auto().Regression(label, useLgbm: false, useSdca: false, useLbfgsPoissonRegression: false));
experiment.SetDataset(train, 5)
.SetRegressionMetric(RegressionMetric.RSquared, label)
.SetPipeline(pipeline)
.SetMaxModelToExplore(1);
var result = await experiment.RunAsync();
result.Metric.Should().BeGreaterThan(0.5);
}
[Fact]
public async Task AutoMLExperiment_Taxi_Fare_CV_5_SamplingKey_Test()
{
var context = new MLContext(1);
var train = DatasetUtil.GetTaxiFareTrainDataView();
var experiment = context.Auto().CreateExperiment();
var label = DatasetUtil.TaxiFareLabel;
var pipeline = context.Auto().Featurizer(train, excludeColumns: new[] { label })
.Append(context.Auto().Regression(label, useLgbm: false, useSdca: false, useLbfgsPoissonRegression: false));
experiment.SetDataset(train, 5, "vendor_id")
.SetRegressionMetric(RegressionMetric.RSquared, label)
.SetPipeline(pipeline)
.SetMaxModelToExplore(1);
var result = await experiment.RunAsync();
result.Metric.Should().BeGreaterThan(0.2);
result.Metric.Should().BeLessThan(0.5);
}
[Fact]
public void AutoMLExperiment_should_use_seed_from_context_if_provided()
{
var context = new MLContext();
var experiment = context.Auto().CreateExperiment();
var settings = experiment.ServiceCollection.BuildServiceProvider().GetRequiredService<AutoMLExperiment.AutoMLExperimentSettings>();
settings.Seed.Should().BeNull();
context = new MLContext(1);
experiment = context.Auto().CreateExperiment();
settings = experiment.ServiceCollection.BuildServiceProvider().GetRequiredService<AutoMLExperiment.AutoMLExperimentSettings>();
settings.Seed.Should().Be(1);
}
}
class DummyTrialRunner : ITrialRunner
{
private readonly int _finishAfterNSeconds;
private readonly IChannel _logger;
public DummyTrialRunner(AutoMLExperiment.AutoMLExperimentSettings automlSettings, int finishAfterNSeconds, IChannel logger)
{
_finishAfterNSeconds = finishAfterNSeconds;
_logger = logger;
}
public void Dispose()
{
}
public async Task<TrialResult> RunAsync(TrialSettings settings, CancellationToken ct)
{
_logger.Info("Update Running Trial");
await Task.Delay(_finishAfterNSeconds * 1000, ct);
ct.ThrowIfCancellationRequested();
_logger.Info("Update Completed Trial");
var metric = 1.000 + 0.01 * settings.TrialId;
return new TrialResult
{
TrialSettings = settings,
DurationInMilliseconds = _finishAfterNSeconds * 1000,
Metric = metric,
Loss = - -metric,
};
}
}
class DummyPeformanceMonitor : IPerformanceMonitor
{
private readonly int _checkIntervalInMilliseconds;
private System.Timers.Timer _timer;
public DummyPeformanceMonitor()
{
_checkIntervalInMilliseconds = 1000;
}
public event EventHandler<TrialPerformanceMetrics> PerformanceMetricsUpdated;
public void Dispose()
{
}
public double? GetPeakCpuUsage()
{
return 100;
}
public double? GetPeakMemoryUsageInMegaByte()
{
return 1000;
}
public void OnPerformanceMetricsUpdatedHandler(TrialSettings trialSettings, TrialPerformanceMetrics metrics, CancellationTokenSource trialCancellationTokenSource)
{
}
public void Start()
{
if (_timer == null)
{
_timer = new System.Timers.Timer(_checkIntervalInMilliseconds);
_timer.Elapsed += (o, e) =>
{
PerformanceMetricsUpdated?.Invoke(this, new TrialPerformanceMetrics() { PeakCpuUsage = 100, PeakMemoryUsage = 1000 });
};
_timer.AutoReset = true;
}
_timer.Enabled = true;
}
public void Pause()
{
_timer.Enabled = false;
}
public void Stop()
{
_timer?.Stop();
_timer?.Dispose();
_timer = null;
}
}
}