//Will work for discrete and continuous values
public override double RunModelForSingleData(double[] data)
{
VerifyDataForRun(data);
double result = _missingValue;
double maxProb = Double.NegativeInfinity;
DistributionNormal dn = new DistributionNormal();
//Do for each target attribute
for (int idx = 0; idx < ClassMeanMatrix.Count; idx++)
{
double prob = 1;
for (int col = 0; col < data.Length; col++)
{
//Multiple probability of each attribute
prob = prob * dn.ProbabilityDensityFunction(data[col],
ClassMeanMatrix[idx][col][0],
ClassSDMatrix[idx][col][0]);
}
if (prob > maxProb)
{
maxProb = prob;
result = targetValues[idx]; //It is important that this index is same as classMeanMatrix and classSDMatrix
}
}
return(result);
}
public void Statistics_Pdf_Example_3()
{
DistributionNormal snd =
new DistributionNormal();
double value = snd.ProbabilityDensityFunction(4, 0, 1);
Assert.IsTrue(SupportFunctions.DoubleCompare(value, .0001));
}