public static void Train(MLContext mlContext)
{
try
{
// STEP 1: Load the data
var trainData = mlContext.Data.LoadFromTextFile(path: "AgeRangeData03_AgeGenderLabelEncodedMoreData.csv",
columns: new[]
{
new TextLoader.Column("Age", DataKind.Single, 0),
new TextLoader.Column("Gender", DataKind.Single, 1)
,
new TextLoader.Column("Label", DataKind.Single, 2)
},
hasHeader: true,
separatorChar: ','
);
var progressHandler = new MulticlassExperimentProgressHandler();
ConsoleHelper.ConsoleWriteHeader("=============== Running AutoML experiment ===============");
Console.WriteLine($"Running AutoML multiclass classification experiment for {ExperimentTime} seconds...");
ExperimentResult <MulticlassClassificationMetrics> experimentResult = mlContext.Auto()
.CreateMulticlassClassificationExperiment(ExperimentTime)
.Execute(trainData, "Label", progressHandler: progressHandler);
// Print top models found by AutoML
Console.WriteLine();
PrintTopModels(experimentResult);
Console.WriteLine();
}
catch (Exception ex)
{
Console.WriteLine(ex);
}
}
public static void AutoTrain()
{
// STEP 1: data loading
IDataView trainingDataView = LoadDataFromCsv(trainingCsv);
IDataView testingDataView = LoadDataFromCsv(testingCsv);
// STEP 2: run an AutoML multiclass classification experiment
WriteLineColor($"{Environment.NewLine}AutoML multiclass classification experiment for {ExperimentTime} seconds...", ConsoleColor.Yellow);
var progressHandler = new MulticlassExperimentProgressHandler();
ExperimentResult <MulticlassClassificationMetrics> experimentResult = Context.Auto()
.CreateMulticlassClassificationExperiment(ExperimentTime)
.Execute(trainingDataView, Label, progressHandler: progressHandler);
// STEP 3: evaluate the model and print metrics
RunDetail <MulticlassClassificationMetrics> bestRun = experimentResult.BestRun;
WriteLineColor($"{Environment.NewLine}Top Trainer (by accuracy)", ConsoleColor.Yellow);
PrintTopModels(experimentResult);
WriteLineColor($"{Environment.NewLine}TRAINING USING: {bestRun.TrainerName}", ConsoleColor.Cyan);
Model = bestRun.Model;
var predictions = Model.Transform(testingDataView);
var metrics = Context.MulticlassClassification.Evaluate(data: predictions, labelColumnName: Label, scoreColumnName: Score, predictedLabelColumnName: PredictedLabel);
PrintMultiClassClassificationMetrics(bestRun.TrainerName, metrics);
// STEP 4: save the model
Context.Model.Save(Model, trainingDataView.Schema, modelPath);
}
public static void Train(MLContext mlContext)
{
try
{
// STEP 1: Load the data
var trainData = mlContext.Data.LoadFromTextFile(path: TrainDataPath,
columns: new[]
{
new TextLoader.Column(nameof(InputData.PixelValues), DataKind.Single, 0, 63),
new TextLoader.Column("Number", DataKind.Single, 64)
},
hasHeader: false,
separatorChar: ','
);
var testData = mlContext.Data.LoadFromTextFile(path: TestDataPath,
columns: new[]
{
new TextLoader.Column(nameof(InputData.PixelValues), DataKind.Single, 0, 63),
new TextLoader.Column("Number", DataKind.Single, 64)
},
hasHeader: false,
separatorChar: ','
);
// STEP 2: Initialize our user-defined progress handler that AutoML will
// invoke after each model it produces and evaluates.
var progressHandler = new MulticlassExperimentProgressHandler();
// STEP 3: Run an AutoML multiclass classification experiment
ConsoleHelper.ConsoleWriteHeader("=============== Running AutoML experiment ===============");
Console.WriteLine($"Running AutoML multiclass classification experiment for {ExperimentTime} seconds...");
ExperimentResult <MulticlassClassificationMetrics> experimentResult = mlContext.Auto()
.CreateMulticlassClassificationExperiment(ExperimentTime)
.Execute(trainData, "Number", progressHandler: progressHandler);
// Print top models found by AutoML
Console.WriteLine();
PrintTopModels(experimentResult);
// STEP 4: Evaluate the model and print metrics
ConsoleHelper.ConsoleWriteHeader("===== Evaluating model's accuracy with test data =====");
RunDetail <MulticlassClassificationMetrics> bestRun = experimentResult.BestRun;
ITransformer trainedModel = bestRun.Model;
var predictions = trainedModel.Transform(testData);
var metrics = mlContext.MulticlassClassification.Evaluate(data: predictions, labelColumnName: "Number", scoreColumnName: "Score");
ConsoleHelper.PrintMulticlassClassificationMetrics(bestRun.TrainerName, metrics);
// STEP 5: Save/persist the trained model to a .ZIP file
mlContext.Model.Save(trainedModel, trainData.Schema, ModelPath);
Console.WriteLine("The model is saved to {0}", ModelPath);
}
catch (Exception ex)
{
Console.WriteLine(ex);
}
}
