ml.net-4.cs

public static void Main(string[] args)
{
    var context = new MLContext();

    var data = GetTrainAndTestData(context, InputDataPath);

    var model = TrainModel(context, data.TrainSet);

    ...
}

private static ITransformer TrainModel(MLContext context, IDataView data)
{
    // Construct training pipeline:
    // - handle missing values
    // - create a column for the output by copying the one we want to predict to the expected name "Label"
    // - create a column for all features using the expected name "Features"
    // - apply a regression
    var pipeline = context.Transforms.CopyColumns(outputColumnName: "Label", inputColumnName: PredictionLabel)
        .Append(context.Transforms.ReplaceMissingValues(
            new MissingValueReplacingEstimator.ColumnOptions(
                "Population", 
                replacementMode: MissingValueReplacingEstimator.ColumnOptions.ReplacementMode.Mean)))
        .Append(context.Transforms.Concatenate("Features", FeatureColumns))
        .Append(context.Regression.Trainers.FastForest());

    var model = pipeline.Fit(data);
    SaveModelAsFile(context, model);
    return model;
}

private static void SaveModelAsFile(MLContext context, ITransformer model)
{
    using (var fileStream = new FileStream(ModelPath, FileMode.Create, FileAccess.Write, FileShare.Write))
    {
        context.Model.Save(model, fileStream);
    }

    Console.WriteLine("The model is saved to {0}", ModelPath);
    Console.WriteLine();
}