public LinearRegression(List <double> _xData, List <double> _yData, double _alpha)
{
xData = _xData;
yData = _yData;
xavg = StatCharacteristicModel.Average.GetAverage(xData);
yavg = StatCharacteristicModel.Average.GetAverage(yData);
alpha = _alpha;
quantil = Quantiles.t_Student(1 - alpha / 2, N - 2);
CalculateScores();
DeterminationCoef = new DetermCoef(this, alpha);
}
public DependendAverageHomogeneition(List <double> firstSequence, List <double> secondSequence)
{
List <double> zSelection = firstSequence.Zip(secondSequence, (x, y) => x - y).ToList();
var N = zSelection.Count;
var zAvg = StatCharacteristicModel.Average.GetAverage(zSelection);
var sigma = NormalDistribution.GetSigmaScore(zSelection);
Statistic = zAvg * Math.Sqrt(N) / sigma;
Statistic = double.IsNaN(Statistic) ? 0 : Statistic;
Quantil = Quantiles.t_Student(1 - alpha / 2, N - 1);
}
public IndependendAverageHomogeneition(List <double> firstSequence, List <double> secondSequence)
{
int N1 = firstSequence.Count;
int N2 = secondSequence.Count;
double d1 = NormalDistribution.GetDispersion(firstSequence);
double d2 = NormalDistribution.GetDispersion(secondSequence);
double avg1 = StatCharacteristicModel.Average.GetAverage(firstSequence);
double avg2 = StatCharacteristicModel.Average.GetAverage(secondSequence);
Quantil = Quantiles.t_Student(1 - alpha / 2, N1 + N2 - 2);
if (N1 + N2 > 25)
{
Statistic = (avg1 - avg2) / Math.Sqrt(d1 / N1 + d2 / N2);
}
else
{
Statistic = ((avg1 - avg2) / Math.Sqrt(((N1 - 1) * d1 + (N2 - 1) * d2) / N1 + N2 - 2)) * Math.Sqrt(N1 * N2 / (N1 + N2));
}
Statistic = double.IsNaN(Statistic) ? 0 : Statistic;
}
public override double GetQuantil()
{
return(Quantiles.t_Student(1 - alpha / 2, N - 2));
}
public ParabolicRegression(List <double> _xData, List <double> _yData, double _alpha)
{
alpha = _alpha;
xData = _xData;
xavg = StatCharacteristicModel.Average.GetAverage(xData);
yData = _yData;
yavg = StatCharacteristicModel.Average.GetAverage(yData);
xDataPow2 = xData.Select(d => d * d).ToList();
xPow2avg = StatCharacteristicModel.Average.GetAverage(xDataPow2);
xDataPow3 = xData.Select(d => d * d * d).ToList();
xPow3avg = StatCharacteristicModel.Average.GetAverage(xDataPow3);
xDataDisp = StatCharacteristicModel.StandartDeviationSkew.GetValue(xData);
xDataDisp *= xDataDisp;
phi0 = 1;
quantil = Quantiles.t_Student(1 - alpha / 2, N - 3);
var a = yavg;
var b = GetB();
var c = GetC();
restDisp = RestDisp(a, b, c);
A = new ParabolicalScoreA(restDisp, N)
{
Name = "a0", Value = a, Quantil = quantil
};
var xSd = StatCharacteristicModel.StandartDeviationSkew.GetValue(xData);
B = new ParabolicalScoreB(restDisp, N, xSd)
{
Name = "a1", Value = b, Quantil = quantil
};
var phi2Pow2Sum = xData.Sum(x => Math.Pow(Phi2(x), 2));
C = new ParabolicalScoreC(restDisp, N, phi2Pow2Sum)
{
Name = "a2", Value = c, Quantil = quantil
};
A.Dispersion = restDisp / N;
B.Dispersion = restDisp / (N * xDataDisp);
C.Dispersion = restDisp / phi2Pow2Sum;
var regressionY = xData.Select(xi => Calculate(xi)).ToList();
regressionYavg = StatCharacteristicModel.Average.GetAverage(regressionY);
Scores = new List <RegressionScore>();
Scores.Add(A);
Scores.Add(B);
Scores.Add(C);
DeterminationCoef = new DetermCoef(this, alpha);
}
