public static void Main()
{
string path = "data/allSamples/";
List <BirdSample> trainingData = buildTrainingData(path);
List <BirdSample> testingData = buildTestingData(path);
BirdSample b = createSampleFromFile(1, path + "mal1_song_8.wav");
using (StreamWriter writer = new StreamWriter("out.csv"))
{
for (int i = 0; i < b.features.Count; i++)
{
writer.Write(b.features[i]);
if (i < b.features.Count - 1)
{
writer.WriteLine(",");
}
}
}
int k = getIdealK(trainingData);
Console.WriteLine("1NN: " + KNN(trainingData, testingData, 1) * 100 + "% Accuracy");
Console.WriteLine(k + "NN: " + KNN(trainingData, testingData, k) * 100 + "% Accuracy");
}
public static double tSumDist(BirdSample a, BirdSample b)
{
double distance = 0;
for (int i = 0; i < 100; i++)
{
distance += Math.Abs(a.tSums[i] - b.tSums[i]);
}
return(distance);
}
public static double KLDivergence(BirdSample p, BirdSample q)
{
double divergence = 0;
int limit = (p.features.Count < q.features.Count) ? p.features.Count : q.features.Count;
for (int i = 0; i < limit; i++)
{
divergence += p.features[i] * Math.Log(p.features[i] / q.features[i], 2);
}
return(divergence);
}
public static double sumDist(BirdSample a, BirdSample b)
{
double distance = 0;
for (int i = 0; i < 100; i++)
{
double d = a.sums[i] - b.sums[i];
distance += Math.Abs(d);
}
return(distance);
}
public static double birdDist(BirdSample a, BirdSample b)
{
double distance = 0;
int limit = (a.features.Count < b.features.Count) ? a.features.Count : b.features.Count;
for (int i = 0; i < limit; i++)
{
distance += Math.Abs(a.features[i] - b.features[i]);
}
//Average limit, if it's absolute the shortest sample will always be predicted.
return(distance / limit);
}
public static BirdSample createSampleFromFile(int label, string filename)
{
BirdSample bs = new BirdSample();
List <double> bytes = convertToDecimal(filename);
bs.label = label;
bs.features = bytes;
//bs.normalizeFeatures();
bs.isolatePeaks(0);
//bs.normalizeZeroToOne();
bs.calculateSums();
bs.calculateTSums();
return(bs);
}
public static double crossCorrelation(BirdSample a, BirdSample b)
{
double similarity_a_b = 0;
double distance = 0;
int limit = (a.features.Count < b.features.Count) ? a.features.Count : b.features.Count;
for (int i = 0; i < limit; i++)
{
//distance += Math.Pow(a.features[i], 2) - a.features[i]*b.features[i];
similarity_a_b = checked (similarity_a_b + a.features[i] * b.features[i]);
}
//Console.WriteLine("mal5_song1 vs mal5_song2: {0}", similarity_a_b);
return(similarity_a_b);
}