public double InferEvidence(double[] trainingData)
{
double logEvidence;
ModelDataMixed posteriors = base.InferModelData(trainingData);
logEvidence = InferenceEngine.Infer <Bernoulli>(Evidence).LogOdds;
return(logEvidence);
}
public static void RunCyclingTime3()
{
ModelDataMixed initPriors;
double[] trainingData =
new double[] { 13, 17, 16, 12, 13, 12, 14, 18, 16, 16, 27, 32 };
initPriors.AverageTimeDist = new Gaussian[] {
new Gaussian(15.0, 100.0), //O
new Gaussian(30.0, 100.0) //E
};
initPriors.TrafficNoiseDist = new Gamma[] {
new Gamma(2.0, 0.5), //O
new Gamma(2.0, 0.5) //E
};
initPriors.MixingDist = new Dirichlet(1, 1);
CyclistMixedTraining cyclistMixedTraining = new CyclistMixedTraining();
cyclistMixedTraining.CreateModel();
cyclistMixedTraining.SetModelData(initPriors);
ModelDataMixed posteriors = cyclistMixedTraining.InferModelData(trainingData);
//Print results
Console.WriteLine("Average time distribution 1 = {0:f2}", posteriors.AverageTimeDist[0]);
Console.WriteLine("Average time distribution 2 = {0:f2}", posteriors.AverageTimeDist[1]);
Console.WriteLine("Noise distribution 1 = {0:f2}", posteriors.TrafficNoiseDist[0]);
Console.WriteLine("Noise distribution 2 = {0:f2}", posteriors.TrafficNoiseDist[1]);
Console.WriteLine("Mixing coefficient distribution = {0:f2}", posteriors.MixingDist);
CyclistMixedPrediction cyclistMixedPrediction = new CyclistMixedPrediction();
cyclistMixedPrediction.CreateModel();
cyclistMixedPrediction.SetModelData(posteriors);
Gaussian tomorrowsTime = cyclistMixedPrediction.InferTomorrowsTime();
double tomorrowsMean = tomorrowsTime.GetMean();
double tomorrowsStdDev = Math.Sqrt(tomorrowsTime.GetVariance());
//Print results
Console.WriteLine("Tomorrows average time: {0:f2}", tomorrowsMean);
Console.WriteLine("Tomorrows standard deviation: {0:f2}", tomorrowsStdDev);
Console.WriteLine(
"Probability that tomorrow's time is < 18 min: {0}",
cyclistMixedPrediction.InferTomorrowsTime()
);
}
public virtual void SetModelData(ModelDataMixed modelData)
{
AverageTimePriors.ObservedValue = modelData.AverageTimeDist;
TrafficNoisePriors.ObservedValue = modelData.TrafficNoiseDist;
MixingPrior.ObservedValue = modelData.MixingDist;
}