public void Dataset(
string[] labels,
DatasetName datasetName,
object[,] x,
object[] y)
{
if (x == null ||
y == null ||
x.Rank != 2 ||
x.GetLength(0) == 0 ||
y.Length == 0)
{
TraceListeners.Log(TraceEventType.Error, 0,
"Invalid number of records in the " + datasetName.ToString() + " dataset!", true, true);
}
if (x.GetLength(1) != y.Count())
{
TraceListeners.Log(TraceEventType.Error, 0,
"Number of observations in features and target do not match!", true, true);
}
if (y.All(v => (string)v != labels[0]) ||
y.All(v => (string)v != labels[1]))
{
TraceListeners.Log(TraceEventType.Error, 0,
"All the labels must be present in the " + datasetName.ToString() + " dataset!", true, true);
}
if (x.GetLength(1) < x.GetLength(0))
{
TraceListeners.Log(TraceEventType.Error, 0,
"Number of observations is less than number of features!", true, true);
}
}
public override void InitialiseFeatures(
DatasetName datasetName,
object[,] x,
object[] y)
{
Debug.Assert(x != null, "The feature vector must not be null.");
Debug.Assert(y != null, "The targe variable must not be null.");
Vector[] xV = null;
Range r = null;
// Validate
_validate.Dataset(
labels: _mapping.GetClassLabels().ToArray(),
datasetName: datasetName,
x: x,
y: y);
// Set meta data
_numFeatures = x.GetLength(0);
_numObservations = x.GetLength(1);
// Transpose
double[][] xTrans = new double[_numObservations][];
for (int j = 0; j < _numObservations; j++)
{
xTrans[j] = new double[_numFeatures];
for (int i = 0; i < _numFeatures; i++)
{
xTrans[j][i] = (double)x[i, j];
}
}
// Set target
_y = Variable.Observed(
Array.ConvertAll(y, v => (
System.Convert.ToInt64(v) > 0) ? true : false)).Named(
"y." + datasetName.ToString());
// Set features
xV = new Vector[_numObservations];
r = _y.Range.Named("person");
for (int i = 0; i < _numObservations; i++)
{
xV[i] = Vector.FromArray(xTrans[i]);
}
_x = Variable.Observed(xV, r).Named("x." + datasetName.ToString());
_availableDatasetName = datasetName;
}
public override void Predict(
string inputModelFileName,
DistributionName distributionName,
InferenceAlgorithm inferenceEngineAlgorithm,
double noise)
{
// Validate
// _validate.Predict(inputModelFileName);
// Initialise
List <IDictionary <string, double> > yPredicDistrib = new List <IDictionary <string, double> >();
List <string> yPredicLabel = new List <string>();
VariableArray <bool> yTest = Variable.Array <bool>(new Range(_numObservations)).
Named("y." + _availableDatasetName.ToString()); // to do: correct the size
Bernoulli[] yPredic;
// The inference engine
_engine[DistributionType.Posterior] = SetInferenceEngine(
inferenceEngineAlgorithm, 1);
// Infer postrior weights from training
_w[DistributionType.Posterior] = InferWeights(
distributionType: DistributionType.Posterior,
distributionName: distributionName,
hyperParameters: null);
// BPM
BayesPointMachine(_x, yTest, _w[DistributionType.Posterior], noise);
// predict
// _yPredicLabel = _engine[DistributionType.Posterior].Infer(_y);
yPredic = _engine[DistributionType.Posterior].Infer <Bernoulli[]>(yTest);
for (int i = 0; i < yPredic.Length; i++)
{
yPredicDistrib.Add(new Dictionary <string, double>()
{
{ Convert.ToInt32(yPredic[i].GetProbFalse() > _cutoffPoint).ToString(), yPredic[i].GetProbFalse() },
{ Convert.ToInt32(yPredic[i].GetProbTrue() > _cutoffPoint).ToString(), yPredic[i].GetProbTrue() }
});
yPredicLabel.Add(Convert.ToInt32(yPredic[i].GetProbTrue() > _cutoffPoint).ToString());
}
_yPredicDistrib = yPredicDistrib;
_yPredicLabel = yPredicLabel;
}
public object[] GetTargetConverted(DatasetName datasetName)
{
// Evaluate
if (!_targetConverted.ContainsKey(datasetName))
{
TraceListeners.Log(TraceEventType.Error, 0,
"Dataset does not exist: " + datasetName.ToString(), true, true);
}
// Get
return(_targetConverted[datasetName]);
}
public string[,] GetDataset(DatasetName datasetName)
{
// Evaluate
if (!_datasets.ContainsKey(datasetName))
{
TraceListeners.Log(TraceEventType.Error, 0,
"Dataset does not exist: " + datasetName.ToString(), true, true);
}
// Get
return(_datasets[datasetName]);
}
public void Set(
Dictionary <string, List <string> > dataset,
DatasetName datasetName,
string[] featuresByOrder)
{
TraceListeners.Log(TraceEventType.Information, 0, "Features::Set...", false, true);
int numFeatures = dataset.Keys.Count;
int numObservations = dataset[dataset.Keys.First()].Count;
// Evaluate
if (_datasets.ContainsKey(datasetName))
{
TraceListeners.Log(TraceEventType.Warning, 0,
"Overwriting: " + datasetName.ToString(), false, true);
}
// Set
try
{
_featuresKey[datasetName] = new Dictionary <string, int>();
_datasets[datasetName] = new string[numFeatures, numObservations];
int i = 0;
foreach (string k in featuresByOrder)
{
_featuresKey[datasetName][k] = i;
for (int j = 0; j < numObservations; j++)
{
_datasets[datasetName][i, j] = dataset[k][j];
}
i++;
}
}
catch (Exception e)
{
TraceListeners.Log(TraceEventType.Error, 0, e.ToString(), true, true);
}
}
public void ConvertFeatures(
DatasetName datasetName,
bool standardConversion = false)
{
TraceListeners.Log(TraceEventType.Information, 0,
"Features::ConvertFeatures...", false, true);
// Evaluate
if (!_datasets.ContainsKey(datasetName))
{
TraceListeners.Log(TraceEventType.Error, 0,
"Dataset does not exist: " + datasetName.ToString(), true, true);
}
if (_featuresConverted.ContainsKey(datasetName))
{
TraceListeners.Log(TraceEventType.Warning, 0,
"Overwriting: " + datasetName.ToString(), false, true);
}
// Initialise
string featureType = null;
int numFeatures = _datasets[datasetName].GetLength(0) - 1;
int numObservations = _datasets[datasetName].GetLength(1);
SetFeaturesMetdata();
// Convert features
featureType = (standardConversion) ? "double" : null;
_featuresConverted[datasetName] = new object[numFeatures, numObservations];
string featureName;
for (int i = 0, iOrig = 0; iOrig < numFeatures + 1; i++, iOrig++)
{
featureName = _featuresKey[datasetName].FirstOrDefault(v => v.Value == iOrig).Key;
// Skip the label
if (featureName == _varsMetadataLabelName)
{
i--;
continue;
}
// Convert observations
TraceListeners.Log(TraceEventType.Information, 0,
" ...Convert: " + featureName + " to " +
featureType + " ...", false, true);
for (int j = 0; j < numObservations; j++)
{
_featuresConverted[datasetName][i, j] =
Convert(datasetName, featureName, iOrig, j, featureType);
}
}
// Convert the label variable to labelType
featureType = (standardConversion) ? "string" : null;
_targetConverted[datasetName] = new object[numObservations];
{
int i = _featuresKey[datasetName][_varsMetadataLabelName];
for (int j = 0; j < numObservations; j++)
{
_targetConverted[datasetName][j] =
Convert(datasetName, _varsMetadataLabelName, i, j, featureType);
}
}
}