如何将带有数据的string [] []转换为IDataView-Object？ （ML.NET）

Question

I actually did all the Microsoft Tutorials for ML.NET and want to build my own models now. I want to convert string[][] data to an IDataView-Object , as I want to use it in an ML.NET-Model for binary classification.

So far I have always used data from external text or CSV files for the training. Now I want to use data stored in string[][] data . In data[0][] are the text values and in data[1][] are the boolean values.

I fail to convert the existing nested array into an IDataView object . I have already tried to use the following code:

 public class BinaryData
        {

            public string Text { get; set; }


            public bool Label { get; set; }
        }

// The data is collected from an Excel-Table with some functions and saved in this nested array: 

string[][] data = form.GetDataSelection().GetDataContainer().textCols;



BinaryData[] inMemoryCollection = new BinaryData[data[0].Length];
            for (int i = 0; i < data[0].Length-1; i++)
            {

                inMemoryCollection[i] = new BinaryData
                {
                    Text = data[0][i],
                    Label = Convert.ToBoolean(Convert.ToInt64(data[1][i]))
                };                             
            } 


IDataView dataView = mlContext.Data.LoadFromEnumerable<BinaryData>(inMemoryCollection);

My implementation is based on the tutorial from Microsoft .

It works until I want to use the Fit()-Method . I get the following error-message:

System.InvalidOperationException: 'Splitter/consolidator worker encountered exception while consuming source data'

I hope somebody can help me out here. Many thanks in advance!

Answer 1

it works with an two-dimensional array [,] . I used the Method from this post to convert the jagged array [][] into a two-dimensional array and changed my code a little bit:

string[][] data_jagged = form.GetDataSelection().GetDataContainer().textCols;
string[,] data = To2D(data_jagged);

BinaryData[] inMemoryCollection = new BinaryData[data_jagged[0].Length];
            for (int i = 0; i < data_jagged[0].Length; i++)
            {

                inMemoryCollection[i] = new BinaryData
                {
                    Text = data[0,i],
                    Label = Convert.ToBoolean(Convert.ToInt64(data[1,i]))
                };                             
            } 


IDataView dataView = mlContext.Data.LoadFromEnumerable<BinaryData>(inMemoryCollection);

Thanks to Eric for his help.