public void TestSrCnnAnomalyDetectorWithSeasonalAnomalyData(
[CombinatorialValues(SrCnnDeseasonalityMode.Stl, SrCnnDeseasonalityMode.Mean, SrCnnDeseasonalityMode.Median)] SrCnnDeseasonalityMode mode
)
{
var ml = new MLContext(1);
IDataView dataView;
var dataPath = GetDataPath("Timeseries", "period_anomaly.csv");
// Load data from file into the dataView
dataView = ml.Data.LoadFromTextFile <TimeSeriesDataDouble>(dataPath, hasHeader: true);
// Setup the detection arguments
string outputColumnName = nameof(SrCnnAnomalyDetection.Prediction);
string inputColumnName = nameof(TimeSeriesDataDouble.Value);
// Do batch anomaly detection
var options = new SrCnnEntireAnomalyDetectorOptions()
{
Threshold = 0.23,
BatchSize = -1,
Sensitivity = 53.0,
DetectMode = SrCnnDetectMode.AnomalyAndMargin,
Period = 288,
DeseasonalityMode = mode
};
var outputDataView = ml.AnomalyDetection.DetectEntireAnomalyBySrCnn(dataView, outputColumnName, inputColumnName, options);
// Getting the data of the newly created column as an IEnumerable of SrCnnAnomalyDetection.
var predictionColumn = ml.Data.CreateEnumerable <SrCnnAnomalyDetection>(
outputDataView, reuseRowObject: false);
var anomalyStartIndex = 2988;
var anomalyEndIndex = 3095;
int k = 0;
foreach (var prediction in predictionColumn)
{
Assert.Equal(7, prediction.Prediction.Length);
if (anomalyStartIndex <= k && k <= anomalyEndIndex)
{
Assert.Equal(1, prediction.Prediction[0]);
}
else
{
Assert.Equal(0, prediction.Prediction[0]);
}
++k;
}
}
public SrCnnEntireModeler(double threshold, double sensitivity, SrCnnDetectMode detectMode, int period, SrCnnDeseasonalityMode deseasonalityMode)
{
_threshold = threshold;
_sensitivity = sensitivity;
_detectMode = detectMode;
_period = period;
_predictArray = new double[_lookaheadWindowSize + 1];
switch (deseasonalityMode)
{
case SrCnnDeseasonalityMode.Stl:
_deseasonalityFunction = new StlDeseasonality();
break;
case SrCnnDeseasonalityMode.Mean:
_deseasonalityFunction = new MeanDeseasonality();
break;
default:
Contracts.Assert(deseasonalityMode == SrCnnDeseasonalityMode.Median);
_deseasonalityFunction = new MedianDeseasonality();
break;
}
}
public Batch(int batchSize, int outputLength, double threshold, double sensitivity, SrCnnDetectMode detectMode, int period, SrCnnDeseasonalityMode deseasonalityMode)
{
_batchSize = batchSize;
_outputLength = outputLength;
if (batchSize == -1)
{
_previousBatch = new List <double>();
_batch = new List <double>();
}
else
{
_previousBatch = new List <double>(batchSize);
_batch = new List <double>(batchSize);
}
_modeler = new SrCnnEntireModeler(threshold, sensitivity, detectMode, period, deseasonalityMode);
}
