ML.net - Bad value at line 1 in column label

问题

I am following the Iris tutorial for ML.Net, I typed out the instructions instead of copy/pasting them so I could learn the API better, but now I am getting some errors.

When I run this line from the tutorial a System.Reflection.TargetInvocationException is thrown:

var model = pipeline.Train<IrisData, IrisPrediction>();

The console errors I am getting during runtime are:

  Bad value at line 2 in column Label
  ...
  Bad value at line 8 in column Label
  Suppressing further bad value messages
  ...
Processed 150 rows with 150 bad values and 0 format errors
Warning: Term map for output column 'Label' contains no entries.
Automatically adding a MinMax normalization transform, use 'norm=Warn' or 'norm=No' to turn this behavior off.
Using 2 threads to train.
Automatically choosing a check frequency of 2.
  Bad value at line 1 in column Label
  ...
  Suppressing further bad value messages
Processed 150 rows with 150 bad values and 0 format errors
Warning: Skipped 150 instances with missing features/label during training

Here is my IrisData class:

namespace Ronald.A.Fisher
{
    public class IrisData
    {
        [Column("0")]
        public float SepalLength;
        [Column("1")]
        public float SepalWidth;
        [Column("2")]
        public float PetalLength;
        [Column("3")]
        public float PetalWidth;
        [Column("4")]
        [ColumnName("Label")]
        public float Label;
    }

回答1:

After looking at it for a short while I realized that one of my columns had the incorrect data type.

In the class used to load the learning data, IrisData, I used the incorrect data type for Label. Hence the console message: Bad value at line 1 in column Label.

To fix this, I changed the data type for the Label field from float to string:

public class IrisData
{
    ...
    [ColumnName("Label")]
    public string Label;
}

来源：https://stackoverflow.com/questions/50244255/ml-net-bad-value-at-line-1-in-column-label

标签

machine-learning

ml.net