public override double[] GetPriorProbabilities(OptimizationParameterList discreteParameters)
{
const double eps = 0.0001;
double[] priors = new double[NonMissingClassCount];
try
{
EigenPair eig = LinearAlgebra.ComputeSparseEigenPair(LinearAlgebra.Transpose(GetTransitionProbabilityMatrix(discreteParameters, 1)));
ComplexNumber[] eigenValues = eig.EigenValues;
for (int i = 0; i < 4; i++)
{
if (ComplexNumber.ApproxEqual(eigenValues[i], 1, eps))
{
priors = LinearAlgebra.Abs(LinearAlgebra.ComplexToDouble(LinearAlgebra.Transpose(eig.EigenVectors)[i]));
break;
}
}
priors = LinearAlgebra.Normalize(priors);
}
catch (Exception e)
{
throw new NotComputableException("Problem computing the prior: " + e.Message);
}
return(priors);
}
public override EvaluationResults EvaluateModelOnData(Converter <Leaf, SufficientStatistics> predMap, Converter <Leaf, SufficientStatistics> targMap)
{
TwoByTwo fishers2by2 = TwoByTwo.GetInstance(
SufficientStatisticsMapToIntDictionaryMap(predMap, _fullLeafCollection),
SufficientStatisticsMapToIntDictionaryMap(targMap, _fullLeafCollection));
int[] fisherCounts = fishers2by2.ToOneDArray();
List <Score> nullScores;
Score altScore;
ComputeIidScores(fisherCounts, out nullScores, out altScore);
EvaluationResultsFisher results = EvaluationResultsFisher.GetInstance(this, nullScores, altScore, fishers2by2);
#if DEBUG
EvaluationResults results2 = EvaluateModelOnDataGivenParams(predMap, targMap, results);
double eps = 1E-14;
Debug.Assert(ComplexNumber.ApproxEqual(results.AltLL, results2.AltLL, eps));
Debug.Assert(ComplexNumber.ApproxEqual(results.NullLL, results2.NullLL, eps));
Debug.Assert(ComplexNumber.ApproxEqual(results.ComputePValue(), results2.ComputePValue(), eps));
#endif
return(results);
}
public override EvaluationResults EvaluateModelOnData(Converter <Leaf, SufficientStatistics> predictorMap, Converter <Leaf, SufficientStatistics> targetMap)
{
EvaluationResults evalResults;
int[] fisherCounts = ModelScorer.PhyloTree.FisherCounts(predictorMap, targetMap);
int[] realFisherCounts = fisherCounts; // for compatability when NAIVE_EQUILIBRIUM is set
#if NAIVE_EQUILIBRIUM
//USE THIS FOR BACKWARDS COMPATABILITY
int[] tempCounts = ModelScorer.PhyloTree.CountsOfLeaves(targetMap);
fisherCounts = tempCounts;
#endif
//MessageInitializerDiscrete nullMessageInitializer = MessageInitializerDiscrete.GetInstance(predictorMap, targetMap, NullDistn, fisherCounts, ModelScorer.PhyloTree.LeafCollection);
//if (TryShortCutFromCounts(realFisherCounts, nullMessageInitializer, out evalResults))
//{
// return evalResults;
//}
//Score nullScore = ModelScorer.MaximizeLikelihood(nullMessageInitializer);
bool isInvariant;
Score nullScoreTarg = ComputeSingleVariableScore(predictorMap, targetMap, NullDistn, fisherCounts, out isInvariant);
Score altScore = ComputeConditionalVariableScore(predictorMap, targetMap, nullScoreTarg, fisherCounts);
//(realFisherCounts, nullScoreTarg, out evalResults))
//{
// return evalResults;
//}
//MessageInitializerDiscrete altMessageInitializer = MessageInitializerDiscrete.GetInstance(predictorMap, targetMap, (DistributionDiscreteConditional)AltDistn, nullScore.OptimizationParameters, ModelScorer.PhyloTree.LeafCollection);
//Score condScore = ModelScorer.MaximizeLikelihood(altMessageInitializer);
List <Score> nullScores = new List <Score>();
if (_includePredictorInScore)
{
int[] predFisherCounts = new int[] { realFisherCounts[0], realFisherCounts[2], realFisherCounts[1], realFisherCounts[3] };
Score predNullScore = ComputeSingleVariableScore(targetMap, predictorMap, NullDistn, predFisherCounts, out isInvariant);
nullScores.Add(predNullScore);
// conditional model altScore doesn't include predLL. If we're here, we want to add it to make it comparable to joint or reverseConditional
altScore = Score.GetInstance(altScore.Loglikelihood + predNullScore.Loglikelihood, altScore.OptimizationParameters, altScore.Distribution);
}
nullScores.Add(nullScoreTarg);
evalResults = EvaluationResultsDiscrete.GetInstance(this, nullScores, altScore, realFisherCounts, ChiSquareDegreesOfFreedom);
#if DEBUG
MessageInitializerDiscrete nullMessageInitializer = MessageInitializerDiscrete.GetInstance(predictorMap, targetMap, NullDistn, fisherCounts, ModelScorer.PhyloTree.LeafCollection);
MessageInitializerDiscrete altMessageInitializer = MessageInitializerDiscrete.GetInstance(predictorMap, targetMap, (DistributionDiscreteConditional)AltDistn, nullScoreTarg.OptimizationParameters, ModelScorer.PhyloTree.LeafCollection);
double nullLL = ModelScorer.ComputeLogLikelihoodModelGivenData(nullMessageInitializer, nullScoreTarg.OptimizationParameters);
double altLL = ModelScorer.ComputeLogLikelihoodModelGivenData(altMessageInitializer, altScore.OptimizationParameters);
if (_includePredictorInScore)
{
int[] predFisherCounts = new int[] { realFisherCounts[0], realFisherCounts[2], realFisherCounts[1], realFisherCounts[3] };
MessageInitializerDiscrete nullMessageInitializerPred = MessageInitializerDiscrete.GetInstance(targetMap, predictorMap, NullDistn, predFisherCounts, ModelScorer.PhyloTree.LeafCollection);
double nullLLPred = ModelScorer.ComputeLogLikelihoodModelGivenData(nullMessageInitializerPred, nullScores[0].OptimizationParameters);
altLL += nullLLPred;
}
EvaluationResults evalResults2 = EvaluateModelOnDataGivenParams(predictorMap, targetMap, evalResults);
double eps = 1E-10;
Debug.Assert(ComplexNumber.ApproxEqual(nullLL, nullScoreTarg.Loglikelihood, eps));
Debug.Assert(ComplexNumber.ApproxEqual(altLL, altScore.Loglikelihood, eps));
Debug.Assert(ComplexNumber.ApproxEqual(evalResults.NullLL, evalResults2.NullLL, eps) && ComplexNumber.ApproxEqual(evalResults.AltLL, evalResults2.AltLL, eps), "In ModelEvaluatorCond, results of maximizing LL and computing LL from same params are not the same.");
#endif
return(evalResults);
}
public override EvaluationResults EvaluateModelOnData(Converter <Leaf, SufficientStatistics> v1, Converter <Leaf, SufficientStatistics> v2)
{
List <Leaf> nonMissingLeaves = new List <Leaf>(100);
int seed = 0;
foreach (Leaf leaf in ModelScorer.PhyloTree.LeafCollection)
{
SufficientStatistics class1 = v1(leaf);
SufficientStatistics class2 = v2(leaf);
if (!class1.IsMissing() && !class2.IsMissing())
{
nonMissingLeaves.Add(leaf);
seed ^= (leaf.CaseName + class1.ToString() + class2.ToString()).GetHashCode();
}
}
Random rand = new Random(seed);
nonMissingLeaves = SpecialFunctions.Shuffle(nonMissingLeaves, ref rand);
int groupSize = nonMissingLeaves.Count / _crossValidateCount;
EvaluationResultsCrossValidate combinedResults = null;
double testAltLLSum = 0; // for debugging
double testNullLLSum = 0; // for debugging
for (int i = 0; i < _crossValidateCount; i++)
{
int testStart = i * groupSize;
int trainStart = testStart + groupSize;
Set <Leaf> trainSet = new Set <Leaf>(SpecialFunctions.SubList(nonMissingLeaves, trainStart, nonMissingLeaves.Count - trainStart));
trainSet.AddNewRange(SpecialFunctions.SubList(nonMissingLeaves, 0, testStart));
Converter <Leaf, SufficientStatistics> v1Train = CreateFilteredMap(v1, trainSet);
Converter <Leaf, SufficientStatistics> v2Train = CreateFilteredMap(v2, trainSet);
EvaluationResults trainingResults = InternalEvaluator.EvaluateModelOnData(v1Train, v2Train);
EvaluationResults testAndTrainResult = InternalEvaluator.EvaluateModelOnDataGivenParams(v1, v2, trainingResults);
EvaluationResultsTestGivenTrain testGivenTrainResult = EvaluationResultsTestGivenTrain.GetInstance(this, trainingResults, testAndTrainResult);
if (combinedResults == null)
{
combinedResults = EvaluationResultsCrossValidate.GetInstance(this, testGivenTrainResult);
}
else
{
combinedResults = combinedResults.AddNewResults(testGivenTrainResult);
}
if (double.IsInfinity(combinedResults.AltLL)) // no point in continuing...infinity will kill everything.
{
break;
}
#if DEBUG
double eps = 1E-10;
EvaluationResults testTrainingResults = InternalEvaluator.EvaluateModelOnDataGivenParams(v1Train, v2Train, trainingResults);
Debug.Assert(ComplexNumber.ApproxEqual(testTrainingResults.AltLL, trainingResults.AltLL, eps) &&
ComplexNumber.ApproxEqual(testTrainingResults.NullLL, trainingResults.NullLL, eps));
//Debug.Assert(testTrainingResults.Equals(trainingResults));
double newNullLL = testAndTrainResult.NullLL - trainingResults.NullLL;
double newAltLL = testAndTrainResult.AltLL - trainingResults.AltLL;
Debug.Assert(ComplexNumber.ApproxEqual(newNullLL, testGivenTrainResult.NullLL, eps));
Debug.Assert(ComplexNumber.ApproxEqual(newAltLL, testGivenTrainResult.AltLL, eps));
testNullLLSum += newNullLL;
testAltLLSum += newAltLL;
Debug.Assert(ComplexNumber.ApproxEqual(testNullLLSum, combinedResults.NullLL, eps), "Combined result has wrong NullLL");
Debug.Assert(ComplexNumber.ApproxEqual(testAltLLSum, combinedResults.AltLL, eps), "Combined result has wrong AltLL");
#endif
}
return(combinedResults);
}