/// <summary>
/// Считает коэфициенты регрессии для "У" набором данных "Х"
/// </summary>
/// <param name="x">Набор данных Х</param>
/// <param name="y">К чему стремимся</param>
/// <param name="power">Степенной ряд, для которого делаем счет</param>
/// <returns></returns>
public static double[][] Regression( pmcore.Collection y, pmcore.Collection x, int[] powers)
{
if ( x == null ) throw new Exception("(x) is null");
if ( y == null ) throw new Exception("(y) is null");
if ( powers == null ) throw new Exception("(powers) is null");
if ( powers.Length >= y[0].Count ) throw new Exception("Length is small");
int ValuesCount = y[0].Count;
int xParamCount = x.VectorsCount;
int PolinomLevel = powers.Length;
double [][]result = new double[xParamCount][];
for ( int i = 0; i < xParamCount; i++) result[i] = new double[PolinomLevel];
pmcore.Types.Matrix XMat = new pmcore.Types.Matrix(ValuesCount, PolinomLevel * xParamCount);
pmcore.Types.Matrix YMat = new pmcore.Types.Matrix(ValuesCount, 1);
pmcore.Types.Matrix K = new Types.Matrix();
for ( int i = 0; i < ValuesCount; i++ ) YMat[i][0] = y[0][i];
for ( int i = 0; i < ValuesCount; i++)
for ( int z = 0; z < xParamCount; z++)
for ( int k = 0; k < PolinomLevel; k++)
XMat[i][k + z*(PolinomLevel)] = System.Math.Pow(x[z][i], powers[k]);
K = (XMat.Transpose() * XMat).Inverse()*XMat.Transpose()*YMat;
// ---------- K.Rows
for ( int i = 0; i < xParamCount; i++)
for ( int k = 0; k < PolinomLevel; k++ )
result[i][k] = K[k + i*PolinomLevel][0];
return result;
}
static void Main(string[] args)
{
int t1 = System.Environment.TickCount;
// Regression testing
//pmcore.Collection base_collection = new pmcore.Collection("E:\\.PersonalProjects\\CSharp\\.Library\\.Data\\01-0078_(810).exp(001)_s_pr1.txt", null);
pmcore.Collection base_collection = new pmcore.Collection(2);
base_collection.Add(0,0); base_collection.Add(1,0);
base_collection.Add(0,1); base_collection.Add(1,0.5);
base_collection.Add(0,2); base_collection.Add(1,1);
pmcore.Collection c1 = base_collection.get_SubCollection(new int[]{0});
pmcore.Collection c2 = base_collection.get_SubCollection(new int[]{1});
int []power_array = new int[]{0,1};
double [][]res = pmcore.Math.Regression(c1, c2, power_array);
for ( int i = 0 ; i < c2.VectorsCount; i++)
{
System.Console.WriteLine("");
for ( int k = 0; k < power_array.Length; k++ )
System.Console.Write(res[i][k].ToString("f9") + "\t");
}
System.Console.WriteLine("");
System.Console.WriteLine("");
// Matrix testing
double []in_array = {1, 2, 1,
4, 1, 6,
1, 2, 3};
pmcore.Types.Matrix m1 = new pmcore.Types.Matrix(3, 3, in_array);
pmcore.Types.Matrix m2 = new pmcore.Types.Matrix(m1);
pmcore.Types.Matrix m3 = m1+m2;
pmcore.Types.Matrix m4 = (m1+m2)*(m1+m2);
pmcore.Types.Matrix m5 = ((m1+m2)*(m1+m2)).Inverse();
pmcore.Types.Matrix m6 = m4*m5;
//Console.WriteLine(m1.ToString());
//Console.WriteLine(m2.ToString());
Console.WriteLine(m3.ToString());
Console.WriteLine(m4.ToString());
Console.WriteLine(m5.ToString());
Console.WriteLine(m6.ToString());
// Collection testing
pmcore.Collection cc = new pmcore.Collection(2);
cc.set_VectorName(0, "test1");
cc.set_VectorName(1, "test2");
cc.set_VectorVisibility(1, false);
cc.Add(0, 1); cc.Add(1, 11);
cc.Add(0, 2); cc.Add(1, 22);
cc.Add(0, 3); cc.Add(1, 33);
cc.Add(0, 4); cc.Add(1, 44);
//cc.Load_TextFile("E:\\.PersonalProjects\\CSharp\\.Library\\.Data\\01-0078_(810).exp(001)_s_pr1.txt", null);
cc.Save_TextFile("E:\\.PersonalProjects\\CSharp\\.Library\\.Data\\01-0078_(810).exp(002)_s_pr1__.txt", null);
Console.WriteLine(" done");
Console.WriteLine(cc[0][0].ToString());
Console.WriteLine(cc[0][1].ToString());
Console.ReadKey();
}
static void Main(string[] args)
{
int t1 = System.Environment.TickCount;
// Regression testing
//pmcore.Collection base_collection = new pmcore.Collection("E:\\.PersonalProjects\\CSharp\\.Library\\.Data\\01-0078_(810).exp(001)_s_pr1.txt", null);
pmcore.Collection base_collection = new pmcore.Collection(2);
base_collection.Add(0, 0); base_collection.Add(1, 0);
base_collection.Add(0, 1); base_collection.Add(1, 0.5);
base_collection.Add(0, 2); base_collection.Add(1, 1);
pmcore.Collection c1 = base_collection.get_SubCollection(new int[] { 0 });
pmcore.Collection c2 = base_collection.get_SubCollection(new int[] { 1 });
int [] power_array = new int[] { 0, 1 };
double [][] res = pmcore.Math.Regression(c1, c2, power_array);
for (int i = 0; i < c2.VectorsCount; i++)
{
System.Console.WriteLine("");
for (int k = 0; k < power_array.Length; k++)
{
System.Console.Write(res[i][k].ToString("f9") + "\t");
}
}
System.Console.WriteLine("");
System.Console.WriteLine("");
// Matrix testing
double [] in_array = { 1, 2, 1,
4, 1, 6,
1, 2, 3 };
pmcore.Types.Matrix m1 = new pmcore.Types.Matrix(3, 3, in_array);
pmcore.Types.Matrix m2 = new pmcore.Types.Matrix(m1);
pmcore.Types.Matrix m3 = m1 + m2;
pmcore.Types.Matrix m4 = (m1 + m2) * (m1 + m2);
pmcore.Types.Matrix m5 = ((m1 + m2) * (m1 + m2)).Inverse();
pmcore.Types.Matrix m6 = m4 * m5;
//Console.WriteLine(m1.ToString());
//Console.WriteLine(m2.ToString());
Console.WriteLine(m3.ToString());
Console.WriteLine(m4.ToString());
Console.WriteLine(m5.ToString());
Console.WriteLine(m6.ToString());
// Collection testing
pmcore.Collection cc = new pmcore.Collection(2);
cc.set_VectorName(0, "test1");
cc.set_VectorName(1, "test2");
cc.set_VectorVisibility(1, false);
cc.Add(0, 1); cc.Add(1, 11);
cc.Add(0, 2); cc.Add(1, 22);
cc.Add(0, 3); cc.Add(1, 33);
cc.Add(0, 4); cc.Add(1, 44);
//cc.Load_TextFile("E:\\.PersonalProjects\\CSharp\\.Library\\.Data\\01-0078_(810).exp(001)_s_pr1.txt", null);
cc.Save_TextFile("E:\\.PersonalProjects\\CSharp\\.Library\\.Data\\01-0078_(810).exp(002)_s_pr1__.txt", null);
Console.WriteLine(" done");
Console.WriteLine(cc[0][0].ToString());
Console.WriteLine(cc[0][1].ToString());
Console.ReadKey();
}
/// <summary>
/// Считает коэфициенты регрессии для "У" набором данных "Х"
/// </summary>
/// <param name="x">Набор данных Х</param>
/// <param name="y">К чему стремимся</param>
/// <param name="power">Степенной ряд, для которого делаем счет</param>
/// <returns></returns>
public static double [][] Regression(pmcore.Collection y, pmcore.Collection x, int[] powers)
{
if (x == null)
{
throw new Exception("(x) is null");
}
if (y == null)
{
throw new Exception("(y) is null");
}
if (powers == null)
{
throw new Exception("(powers) is null");
}
if (powers.Length >= y[0].Count)
{
throw new Exception("Length is small");
}
int ValuesCount = y[0].Count;
int xParamCount = x.VectorsCount;
int PolinomLevel = powers.Length;
double [][] result = new double[xParamCount][];
for (int i = 0; i < xParamCount; i++)
{
result[i] = new double[PolinomLevel];
}
pmcore.Types.Matrix XMat = new pmcore.Types.Matrix(ValuesCount, PolinomLevel * xParamCount);
pmcore.Types.Matrix YMat = new pmcore.Types.Matrix(ValuesCount, 1);
pmcore.Types.Matrix K = new Types.Matrix();
for (int i = 0; i < ValuesCount; i++)
{
YMat[i][0] = y[0][i];
}
for (int i = 0; i < ValuesCount; i++)
{
for (int z = 0; z < xParamCount; z++)
{
for (int k = 0; k < PolinomLevel; k++)
{
XMat[i][k + z * (PolinomLevel)] = System.Math.Pow(x[z][i], powers[k]);
}
}
}
K = (XMat.Transpose() * XMat).Inverse() * XMat.Transpose() * YMat;
// ---------- K.Rows
for (int i = 0; i < xParamCount; i++)
{
for (int k = 0; k < PolinomLevel; k++)
{
result[i][k] = K[k + i * PolinomLevel][0];
}
}
return(result);
}
