Кривая AUC-ROC при поиске аномалий ML.Net

MLContext ml = new MLContext();
 
            var dataView = ml.Data.LoadFromEnumerable(data);
            var dataSplit = ml.Data.TrainTestSplit(dataView);
 
            string outputColumnName = nameof(SrCnnAnomalyDetection.Prediction);
            string inputColumnName = nameof(TimeSeriesData.Value);
 
            var odv = ml.Transforms.DetectAnomalyBySrCnn(outputColumnName, inputColumnName, windowSize: 64, backAddWindowSize: 5, lookaheadWindowSize: 5, averagingWindowSize: 3, judgementWindowSize: 21);
var model = odv.Fit(dataSplit.TrainSet);
            var predictions = model.Transform(dataSplit.TrainSet);
 
            var results = ml.Data.CreateEnumerable<SrCnnAnomalyDetection>(predictions, reuseRowObject: false).ToList();
                 
 
public class TimeSeriesData
    {
        public float Value { get; set; }
    }
 
    public class SrCnnAnomalyDetection
    {
        [VectorType]        
        public double[] Prediction { get; set; }
    }

var metrics = ml.AnomalyDetection.Evaluate(dataSplit.TestSet, "Prediction");

var pipeline = ml.Transforms.Concatenate(outputColumnName, inputColumnName).Append(ml.Transforms.CopyColumns("Prediction", nameof(SrCnnAnomalyDetection.Prediction)));
 
            var model = pipeline .Fit(dataSplit.TrainSet);

            List<SeriesData> serdat = new List<SeriesData>();
            int k = 0;
            foreach (var pr in results)
            {
                serdat.Add(new SeriesData(k, data[k].Value, Convert.ToSingle(pr.Prediction[0]), Convert.ToSingle(pr.Prediction[1]), pr.Prediction));
 
                k++;
            }

public class SeriesData
    {
        public int Id { get; set; }
        public double Value { get; set; }
        public Single Anomaly { get; set; }
        public Single Score { get; set; }
        [VectorType]
        public double[] Prediction { get; set; }
}

IDataView dv = ml.Data.LoadFromEnumerable<SeriesData>(serdat);

var metrics = ml.AnomalyDetection.Evaluate(dv, "Anomaly", "Score", "Prediction");

var metrics = ml.AnomalyDetection.Evaluate(dv, labelColumnName: "Anomaly", scoreColumnName: "Score", predictionColumnName: "Prediction");