private static List <List <double> > formFeatureVector(eigens input)
{
List <List <double> > output = new List <List <double> >();
//!! Concern, do eigen values of a greater absolute value while negative supersede positive values?
// Alglib eigenVectors and values are sorted in ascending order, so last 3 entries will be the 3 highest vectors
//eigenVector [dimension, vector]
// [row,col]
// Vector1 Vector2 Vector3 ... Vector 12
// dim 1 dim 1 dim 1
// dim 2 dim 2 dim 2
// dim 3 dim 3 dim 3
// dim 4 dim 4 dim 4
// dim 5 dim 5 dim 5
// ... ... ...
// dim 12 dim 12 dim 12
//this makes the feature vector matrix with 3 rows of one eigenvector per
//it is used like this when translating the data to new dimensions in pca
for (int i = width - 1; i > width - 4; i--)
{
List <double> line = new List <double>();
for (int j = 0; j < width; j++)
{
line.Add(input.eigenVectors[j, i]);
}
output.Add(line);
}
return(output);
}
private static void listToDouble(List <List <adder> > inputMatrix, eigens output)
{
for (int i = 0; i < inputMatrix.Count; i++)
{
for (int j = 0; j < inputMatrix.Count; j++)
{
output.matrix[i, j] = inputMatrix[i][j].Value;
}
}
}
static private void findEigenVectors(List <List <adder> > inputMatrix, eigens output)
{
//create a matrix array from the list<list<double>>
double[] valuesOut;
double[,] vectorsOut;
output.matrix = new double[width, width];
listToDouble(inputMatrix, output);
alglib.smatrixevd(output.matrix, width, 1, false, out valuesOut, out vectorsOut);
output.eigenValues = valuesOut;
output.eigenVectors = vectorsOut;
}
public static void clear() //resets all the data
//simple fix for errors created by opening files.
//could be coded better that data isnt left over from
//previous files
{
size = new int();
width = new int();
means = new List <double>();
zeroData = new List <List <double> >();
zeroDataTransposed = new List <List <double> >();
finalData = new List <List <double> >(); // this will be the new data that is under 3 dimensions and is plot-able
zeroMeans = new List <double>();
zeroVariance = new List <double>();
zeroCoVar = new List <List <adder> >();
zeroEigens = new eigens();
featureVector = new List <List <double> >();
}
