Add ML.NET MLContext.Data.LoadFromTextFile<T>

[jzabroski]

https://docs.microsoft.com/en-us/dotnet/machine-learning/how-to-guides/load-data-ml-net#load-data-from-a-single-file

public class HousingData
{
    [LoadColumn(0)]
    public float Size { get; set; }

    [LoadColumn(1, 3)]
    [VectorType(3)]
    public float[] HistoricalPrices { get; set; }

    [LoadColumn(4)]
    [ColumnName("Label")]
    public float CurrentPrice { get; set; }
}
//Create MLContext
MLContext mlContext = new MLContext();

//Load Data
IDataView data = mlContext.Data.LoadFromTextFile<HousingData>("my-data-file.csv", separatorChar: ',', hasHeader: true);

[jzabroski]

It would be cool to see how Microsoft's machine learning code performs here.

[joelverhagen]

I played around with this one a bit and I agree it would be awesome to compare the performance. However, my current benchmark relies on the parsing library to give each row's data in unmapped form (e.g. string[]). It looks like this library does have that granular of an API -- or at least not available in the public surface area.

This would be a cool library to test if and when I add a higher level "data mapping" benchmark. I'll leave this issue open for that time.

[jzabroski]

Seems do-able to add that benchmark. I'll give it a shot if that's ok with you.

[joelverhagen]

Done with https://github.com/joelverhagen/NCsvPerf/pull/38.