public void Train(double[][] observations, int numberOfIterations, double likelihoodTolerance)
{
if (_initialize)
{
Initialize(observations);
}
if (_pi == null || _transitionProbabilityMatrix == null || _emission == null)
{
throw new ApplicationException("Initialize the model with initial valuesss");
}
var model = HiddenMarkovModelStateFactory.GetState(new ModelCreationParameters <Mixture <IMultivariateDistribution> > {
Pi = _pi, TransitionProbabilityMatrix = _transitionProbabilityMatrix, Emissions = _emission
}); //new HiddenMarkovModelState<Mixture<IMultivariateDistribution>>(_pi, _transitionProbabilityMatrix, _emission);
model.Normalized = Normalized;
var alg = new BaumWelchWeightedMixtureDistribution(Helper.Convert(observations),
TimeSensitiveWeightCalculator.Calculate(7, observations.Length),
model);
var estimatedParameters = alg.Run(numberOfIterations, likelihoodTolerance);
_pi = estimatedParameters.Pi;
_transitionProbabilityMatrix = estimatedParameters.TransitionProbabilityMatrix;
_emission = estimatedParameters.Emission;
Likelihood = estimatedParameters.Likelihood;
}
public void Run_TrainingSetAndModel_TrainedModel()
{
var util = new TestDataUtils();
var observationArray = util.GetSvcData(util.FTSEFilePath, new DateTime(2010, 12, 18), new DateTime(2011, 12, 18));
var observations = Helper.Convert(observationArray);
var observationsWeight = Array.ConvertAll(TimeSensitiveWeightCalculator.Calculate(14, observations.Count), x => (decimal)x);
Initialize(observationArray);
var model = HiddenMarkovModelFactory.GetModel(new ModelCreationParameters <Mixture <IMultivariateDistribution> >()
{
Pi = _pi, TransitionProbabilityMatrix = _tpm, Emissions = _emission
});
model.Normalized = true;
var algo = new BaumWelchWeightedMixtureDistribution(observations, observationsWeight, model);
var trainedModelState = algo.Run(MaxIterationNumber, LikelihoodTolerance);
Assert.AreEqual(1d, Math.Round(trainedModelState.Pi.Sum(), 5));
Assert.AreEqual(1d, Math.Round(trainedModelState.TransitionProbabilityMatrix[0].Sum(), 5));
Assert.AreEqual(1d, Math.Round(trainedModelState.TransitionProbabilityMatrix[1].Sum(), 5));
Assert.AreEqual(1d, Math.Round(trainedModelState.TransitionProbabilityMatrix[2].Sum(), 5));
}
