Feature request: Add support for saving/loading IDataView to/from csv/tsv directly and other pandas-like functionalities

[LittleLittleCloud]

This will be a useful feature when using ML.Net to build pipeline in jupyter notebook, especially in data preprocessing steps.

Other useful functions can be

• IDataView.description -> similar to DataFrame.description, which prints summary information of current dataset.
• IDataView.Head(n) -> print first n rows in nice table format
• IDataView[ColumnName] -> get the column of ColumnName

[antoniovs1029]

Hi, @LittleLittleCloud .

Saving and loading from TSV/CSV is already possible with TextLoader and TextSaver. Look for the TextLoaderSaverCatalog.cs for these APIs. So I'm not sure what's the request on this regard.

As for the other functions, they do sound helpful for Jupyter notebooks, and I think they'd be nice additions to ML.NET. By the way, on the dv[name] request, ML.NET does have an extension method called GetColumn:

https://github.com/dotnet/machinelearning/blob/a76936546f5e269fc1da61f33ab541389e445294/src/Microsoft.ML.Data/Utilities/ColumnCursor.cs#L25-L26

[antoniovs1029]

For the record, if users are looking for pandas-like features to create or read IDataViews compatible with ML.NET, there's the DataFrame project:

https://devblogs.microsoft.com/dotnet/an-introduction-to-dataframe/

https://github.com/dotnet/interactive/blob/main/samples/notebooks/csharp/Samples/HousingML.ipynb

Thanks to @eerhardt for the heads-up about it