public void ValidateStates()
{
var enumerator = ProbabilityMatrix.GetEnumerator();
//every row contains one probability distribution (also the initial distribution)
while (enumerator.MoveNextRow())
{
// for each state there is a row. The sum of all columns in a row should be 1.0
var probability = 0.0;
while (enumerator.MoveNextColumn())
{
if (enumerator.CurrentColumnValue != null)
{
probability += enumerator.CurrentColumnValue.Value.Value;
}
else
{
throw new Exception("Entry must not be null");
}
}
if (!Probability.IsOne(probability, 0.000000001))
{
throw new Exception("Probabilities should sum up to 1");
}
}
}
private int ColumnToState(int state) => state; //Do nothing! Just here to make the algorithms more clear.
// Creating matrix phase
// For the initial distribution the process is
// StartWithInitialDistribution()
// while(transitions to add exist) {
// AddInitialTransition();
// }
// FinishInitialDistribution()
internal void StartWithInitialDistribution()
{
ProbabilityMatrix.SetRow(RowOfInitialStates);
}
public void SealProbabilityMatrix()
{
ProbabilityMatrix.OptimizeAndSeal();
}
internal void FinishDistribution()
{
ProbabilityMatrix.FinishRow();
}
internal void AddTransition(int markovChainState, double probability)
{
ProbabilityMatrix.AddColumnValueToCurrentRow(new SparseDoubleMatrix.ColumnValue(StateToColumn(markovChainState), probability));
Transitions++;
}
// For distribution of a state the process is
// StartWithNewDistribution(markovChainSourceState)
// while(transitions to add exist) {
// AddTransition();
// }
// FinishDistribution()
internal void StartWithNewDistribution(int markovChainSourceState)
{
ProbabilityMatrix.SetRow(StateToRow(markovChainSourceState));
}
internal void AddInitialTransition(int markovChainState, double probability)
{
// initial state probabilities are also saved in the ProbabilityMatrix
ProbabilityMatrix.AddColumnValueToCurrentRow(new SparseDoubleMatrix.ColumnValue(StateToColumn(markovChainState), probability));
InitialTransitions++;
}
private void makePrediction(PointQ currentCoordinate, PointQ previousCoordinate, List <PointQ> allEndPoints)
{
foreach (PointQ endPoint in allEndPoints)
{
Double previous_distance = GetDistance(previousCoordinate, endPoint);
Double current_distance = GetDistance(currentCoordinate, endPoint);
//distance to endpoint is increasing which means that the pedestrian is moving away from this point and therefore moving towards another direction
if (current_distance > previous_distance)
{
//rule out this destination/Path
ProbabilityMatrix pM = new ProbabilityMatrix(endPoint, false);
listProbabilityMatrix.Add(pM);
}
//if distance is decreasing towards a certain endpoint this means that pedestrian is moving near to that endpoint
else
{
ProbabilityMatrix pM = new ProbabilityMatrix(endPoint, true);
listProbabilityMatrix.Add(pM);
}
}
foreach (ProbabilityMatrix pM in listProbabilityMatrix)
{
//Add valid paths for calculating probabilty
if (pM.isValid == true)
{
probabilityCount += 1;
validDestination.Add(pM.allEndPoints);
}
else
{
//Dont include in calculating probabilty
}
}
//Total number of Paths/Routes => 6 in our case
double totalnumberofendpoints = allEndPoints.Count();
// total number of valid endpoints => 3 in our case
double totalnumberofValidDestinations = validDestination.Count();
foreach (PointQ p in validDestination)
{
double probability = (1 / totalnumberofValidDestinations) * 100.00;
Console.WriteLine("The probability of pedestrian reaching the point (" + p.X + ", " + p.Y + ") is :" + probability.ToString());
//Write Probability of Pedestrian of Valid paths in CSV file
using (FileStream fs = new FileStream("Ped.csv", FileMode.Append, FileAccess.Write))
{
StreamWriter sw = new StreamWriter(fs);
sw.Write("The probability of pedestrian reaching the point (" + p.X + ", " + p.Y + ") is :" + probability.ToString());
sw.Close();
}
}
}
