public static bool ChooseKNeighbours(TestVectorAndTrainingVectorsCollection TestAndTrainingPair, int k)
{
TestAndTrainingPair.TrainingReuters = TestAndTrainingPair.TrainingReuters.OrderBy(h => h.HowFar).ToList(); //sorted list
var Kneighbours = TestAndTrainingPair.TrainingReuters.Take(k).ToList(); // take k neighbours
return(ChoosePlace(Kneighbours, TestAndTrainingPair.TestReuter));;
}
public static bool CalculateManhattanMetricForOneTestSet(Reuter testSet, List <Reuter> TrainingVectors, int k)
{
double xn = 0;
double yn = 0;
double underSqrt = 0;
TestVectorAndTrainingVectorsCollection result = new TestVectorAndTrainingVectorsCollection();
for (int i = 0; i < TrainingVectors.Count; i++) //wykonujemy petle dla kazdego wzorca treningowego
{
foreach (var word in testSet.VectorFeatures) //sprawdzamy dla kazdego slowa z wektora testowego czy istnieje takie slowo w wektorze treningowym
{
if (TrainingVectors.ElementAt(i).VectorFeatures.ContainsKey(word.Key))
{
yn = TrainingVectors.ElementAt(i).VectorFeatures[word.Key];
}
else
{
yn = 0;
}
xn = word.Value;
underSqrt += Math.Abs(xn - yn);
}
TrainingVectors.ElementAt(i).HowFar = underSqrt;
result.TestReuter = testSet;
underSqrt = 0;
}
result.TrainingReuters = TrainingVectors;
return(KnnAlgorithm.Calculate(result, k));
}
public static bool Calculate(TestVectorAndTrainingVectorsCollection item, int k)
{
return(ChooseKNeighbours(item, k));
}
