/// <summary>
/// Calculates the x-value boundaries of the area that covers a certain amount
/// of cumulative density function of the passed normal-distribution.
/// </summary>
/// <param name="mean">mean of the passed normal-distribution</param>
/// <param name="stdDev">standard-deviation of the passed normal-distribution</param>
/// <param name="areaCoverage">describes the percentage of area that should be covered below
/// the normal-distribution function. This percentage has to be in the interval ]0,1[</param>
/// <returns>a touple containing the lower and upper boundary x-value of the
/// passed area</returns>
public static (double lowerBoundary, double upperBoundary) CalculateBoundaries(
double mean,
double stdDev,
double areaCoverage)
{
if (areaCoverage <= 0 || areaCoverage >= 1)
{
throw new InvalidOperationException($"{nameof(areaCoverage)} has to be in boundary ]0,1[ but is {areaCoverage}");
}
else if (double.IsNaN(mean) || double.IsNaN(stdDev))
{
return(double.NegativeInfinity, double.PositiveInfinity);
}
else if (stdDev == 0)
{
return(mean, mean);
}
var distribution = new Accord.Statistics.Distributions.Univariate.NormalDistribution(mean, stdDev);
var lowerBoundaryPercentage = (1 - areaCoverage) / 2;
var upperBoundaryPercentage = 1 - lowerBoundaryPercentage;
return(
distribution.InverseDistributionFunction(lowerBoundaryPercentage),
distribution.InverseDistributionFunction(upperBoundaryPercentage));
}
public void TwoSampleZTestConstructorTest()
{
Accord.Math.Tools.SetupGenerator(0);
double[] samples1 = new Accord.Statistics.Distributions.Univariate
.NormalDistribution(29.8, 4.0).Generate(200);
double[] samples2 = new Accord.Statistics.Distributions.Univariate
.NormalDistribution(34.7, 5.0).Generate(250);
TwoSampleZTest actual = new TwoSampleZTest(samples1, samples2);
double mean1 = Accord.Statistics.Tools.Mean(samples1);
double mean2 = Accord.Statistics.Tools.Mean(samples2);
double var1 = Accord.Statistics.Tools.Variance(samples1);
double var2 = Accord.Statistics.Tools.Variance(samples2);
int n1 = samples1.Length;
int n2 = samples2.Length;
TwoSampleZTest expected = new TwoSampleZTest(mean1, var1, n1, mean2, var2, n2);
Assert.AreEqual(expected.EstimatedValue1, actual.EstimatedValue1);
Assert.AreEqual(expected.EstimatedValue2, actual.EstimatedValue2);
Assert.AreEqual(expected.StandardError, actual.StandardError);
Assert.AreEqual(expected.Statistic, actual.Statistic);
Assert.IsTrue(actual.Significant);
}
public void TwoSampleZTestConstructorTest()
{
Accord.Math.Tools.SetupGenerator(0);
double[] samples1 = new Accord.Statistics.Distributions.Univariate
.NormalDistribution(29.8, 4.0).Generate(200);
double[] samples2 = new Accord.Statistics.Distributions.Univariate
.NormalDistribution(34.7, 5.0).Generate(250);
TwoSampleZTest actual = new TwoSampleZTest(samples1, samples2);
double mean1 = Accord.Statistics.Tools.Mean(samples1);
double mean2 = Accord.Statistics.Tools.Mean(samples2);
double var1 = Accord.Statistics.Tools.Variance(samples1);
double var2 = Accord.Statistics.Tools.Variance(samples2);
int n1 = samples1.Length;
int n2 = samples2.Length;
TwoSampleZTest expected = new TwoSampleZTest(mean1, var1, n1, mean2, var2, n2);
Assert.AreEqual(expected.EstimatedValue1, actual.EstimatedValue1);
Assert.AreEqual(expected.EstimatedValue2, actual.EstimatedValue2);
Assert.AreEqual(expected.StandardError, actual.StandardError);
Assert.AreEqual(expected.Statistic, actual.Statistic);
Assert.IsTrue(actual.Significant);
}
private void variableanalysisbutton_Click(object sender, EventArgs e)
{
clear();
if (comboBox1.Text.Equals("Зависимая переменная"))
{
number_cols = 0;
}
else
{
string resultString = string.Join(string.Empty, Regex.Matches(comboBox1.Text, @"\d+").OfType <Match>().Select(m => m.Value));
int.TryParse(resultString, out number_cols);
}
for (int i = 0; i < dataGridView1.RowCount; i++)
{
double val1 = Convert.ToDouble(dataGridView1.Rows[i].Cells[number_cols].Value);
if (!valuesUQ.Contains(val1))
{
valuesUQ.Add(val1);
}
values.Add(val1);
}
Numberofvalues.Text = values.Count.ToString();
numberuniquevalues.Text = valuesUQ.Count.ToString();
scaletype.Text = AnalysisClass.getScale(valuesUQ.Count);
Accord.Statistics.Distributions.Univariate.NormalDistribution normDist =
Accord.Statistics.Distributions.Univariate.NormalDistribution.Standard;
if (!scaletype.Text.Equals("дихотомическая"))
{
Accord.Statistics.Testing.KolmogorovSmirnovTest kstX =
new Accord.Statistics.Testing.KolmogorovSmirnovTest(values.ToArray(), normDist);
TypeDistr.Text = AnalysisClass.getDistributionScale(kstX.Significant);
avg.Text = Math.Round(MathNet.Numerics.Statistics.Statistics.Mean(values), 3).ToString();
standarddeviation.Text = Math.Round(MathNet.Numerics.Statistics.Statistics.PopulationStandardDeviation(values), 3).ToString();
skewness.Text = Math.Round(MathNet.Numerics.Statistics.Statistics.PopulationSkewness(values), 3).ToString();
piko.Text = Math.Round(MathNet.Numerics.Statistics.Statistics.PopulationKurtosis(values), 3).ToString();
plotting.CreateGraphNormDistr(zg1, values, valuesUQ);
}
else
{
TypeDistr.Text = "Невозможно определить определить";
avg.Text = "Невозможно определить определить";
standarddeviation.Text = "Невозможно определить определить";
skewness.Text = "Невозможно определить определить";
piko.Text = "Невозможно определить определить";
zg1.GraphPane.CurveList.Clear();
zg1.GraphPane.GraphObjList.Clear();
zg1.Refresh();
}
}
private void Recalculate(double mean, double stdDev)
{
Accord.DoubleRange range = new Accord.DoubleRange(-100, 100);
Task.Run(() =>
{
Accord.Statistics.Distributions.Univariate.NormalDistribution normalDistribution = new Accord.Statistics.Distributions.Univariate.NormalDistribution(mean, stdDev);
double[] x = Accord.Math.Vector.Range(-100, 100, 0.1);
double[] y = x.Apply(normalDistribution.ProbabilityDensityFunction);
return(pmservice.Create(range, "", x, y, false));
}).ContinueWith(a => this.Dispatcher.InvokeAsync(async() => PlotModel = await a, System.Windows.Threading.DispatcherPriority.Background));
}
/// <summary>
/// Generates a random observation from the passed normal-distribution
/// </summary>
/// <param name="mean">mean of the passed normal-distribution</param>
/// <param name="stdDev">standard-deviation of the passed normal-distribution</param>
/// <returns>the generated random observation</returns>
public static double GenerateObservation(double mean, double stdDev)
{
if (stdDev == 0)
{
return(mean);
}
var distribution = new Accord.Statistics.Distributions.Univariate.NormalDistribution(mean, stdDev);
return(distribution.Generate());
}
//static private decimal GetBlsCallPrice(double S, double K, double T, double sigma, double r)
// {
// Accord.Statistics.Distributions.Univariate.NormalDistribution normDist = new Accord.Statistics.Distributions.Univariate.NormalDistribution();
// return decimal.Ceiling((decimal)(S * normDist.DistributionFunction(D1(S, K, T, sigma, r)) - K * Math.Exp(-r * T) * normDist.DistributionFunction(D2(S, K, T, sigma, r)))*100m)/100m;
// //return S * normDist.DistributionFunction(D1(S, K, T, sigma, r)) - K * Math.Exp(-r * T) * normDist.DistributionFunction(D2(S, K, T, sigma, r));
// }
// static private decimal GetBlsPutPrice(double S, double K, double T, double sigma, double r)
// {
// Accord.Statistics.Distributions.Univariate.NormalDistribution normDist = new Accord.Statistics.Distributions.Univariate.NormalDistribution();
// return decimal.Ceiling((decimal)(-S * normDist.DistributionFunction(-D1(S, K, T, sigma, r)) + K * Math.Exp(-r * T) * normDist.DistributionFunction(-D2(S, K, T, sigma, r)))*100m)/100m;
// //return -S * normDist.DistributionFunction(-D1(S, K, T, sigma, r)) + K * Math.Exp(-r * T) * normDist.DistributionFunction(-D2(S, K, T, sigma, r));
// }
// static public decimal GetBlsPrice(double S, double K, double T, double sigma, double r, char type)
// {
// return type == 'c' ? GetBlsCallPrice(S, K, T, sigma, r) : GetBlsPutPrice(S, K, T, sigma, r);
// }
static public double GetBlsDelta(double S, double K, double T, double sigma, double r, char type)
{
Accord.Statistics.Distributions.Univariate.NormalDistribution normDist = new Accord.Statistics.Distributions.Univariate.NormalDistribution();
return type == 'c'? (normDist.DistributionFunction(D1(S, K, T, sigma, r))) : (normDist.DistributionFunction(D1(S, K, T, sigma, r))-1);
}
private void doAnalysisBtn_Click(object sender, EventArgs e)
{
List <Double> xUniqVals = new List <Double>();
List <Double> yUniqVals = new List <Double>();
for (int i = 0; i < dataGV_data.RowCount - 1; i++)
{
Double val1 = Convert.ToDouble(dataGV_data.Rows[i].Cells[0].Value);
Double val2 = Convert.ToDouble(dataGV_data.Rows[i].Cells[1].Value);
if (!xUniqVals.Contains(val1))
{
xUniqVals.Add(val1);
}
if (!yUniqVals.Contains(val2))
{
yUniqVals.Add(val2);
}
xVals.Add(val1);
yVals.Add(val2);
}
volumeTxtBx.Text = (dataGV_data.RowCount - 1).ToString();
xUniqCountTxtBx.Text = xUniqVals.Count.ToString();
yUniqCountTxtBx.Text = yUniqVals.Count.ToString();
xSchaleTxtBx.Text = AnalysisClass.getScale(xUniqVals.Count);
ySchaleTxtBx.Text = AnalysisClass.getScale(yUniqVals.Count);
Accord.Statistics.Distributions.Univariate.NormalDistribution normDist =
Accord.Statistics.Distributions.Univariate.NormalDistribution.Standard;
cl.CreateGraph(zg1, dataGV_data);
if (!xSchaleTxtBx.Text.Equals("дихотомическая"))
{
Accord.Statistics.Testing.KolmogorovSmirnovTest kstX =
new Accord.Statistics.Testing.KolmogorovSmirnovTest(xVals.ToArray(), normDist);
distrXTxtBx.Text = AnalysisClass.getDistributionScale(kstX.Significant);
meanXTxtBx.Text = MathNet.Numerics.Statistics.Statistics.Mean(xVals).ToString();
stddevXTxtBx.Text = MathNet.Numerics.Statistics.Statistics.PopulationStandardDeviation(xVals).ToString();
skewXTxtBx.Text = MathNet.Numerics.Statistics.Statistics.PopulationSkewness(xVals).ToString();
kurtXTxtBx.Text = MathNet.Numerics.Statistics.Statistics.PopulationKurtosis(xVals).ToString();
cl.CreateGraph3(zg3, dataGV_data);
}
if (!ySchaleTxtBx.Text.Equals("дихотомическая"))
{
Accord.Statistics.Testing.KolmogorovSmirnovTest kstY =
new Accord.Statistics.Testing.KolmogorovSmirnovTest(yVals.ToArray(), normDist);
distrYTxtBx.Text = AnalysisClass.getDistributionScale(kstY.Significant);
meanYTxtBx.Text = MathNet.Numerics.Statistics.Statistics.Mean(yVals).ToString();
stddevYTxtBx.Text = MathNet.Numerics.Statistics.Statistics.PopulationStandardDeviation(yVals).ToString();
skewYTxtBx.Text = MathNet.Numerics.Statistics.Statistics.PopulationSkewness(yVals).ToString();
kurtYTxtBx.Text = MathNet.Numerics.Statistics.Statistics.PopulationKurtosis(yVals).ToString();
cl.CreateGraph2(zg2, dataGV_data);
}
coefficientsTxtBx.Text = AnalysisClass.getCorrelationCoefficients(
xSchaleTxtBx.Text, distrXTxtBx.Text, ySchaleTxtBx.Text, distrYTxtBx.Text, dataGV_data.RowCount - 1);
xVals.Clear();
yVals.Clear();
xUniqVals.Clear();
yUniqVals.Clear();
}
static public decimal GetBlsPutPrice(double S, double K, double T, double sigma, double r)
{
Accord.Statistics.Distributions.Univariate.NormalDistribution normDist = new Accord.Statistics.Distributions.Univariate.NormalDistribution();
return decimal.Round((decimal)(-S * normDist.DistributionFunction(-D1(S, K, T, sigma, r)) + K * Math.Exp(-r * T) * normDist.DistributionFunction(-D2(S, K, T, sigma, r)))+0.01m, 2, MidpointRounding.AwayFromZero);
}
public static double zScore(double alpha=0.05)
{
Accord.Statistics.Distributions.Univariate.NormalDistribution nd = new Accord.Statistics.Distributions.Univariate.NormalDistribution();
return nd.InverseDistributionFunction(1 - (alpha / 2));
}
public static double zScore(double alpha = 0.05)
{
Accord.Statistics.Distributions.Univariate.NormalDistribution nd = new Accord.Statistics.Distributions.Univariate.NormalDistribution();
return(nd.InverseDistributionFunction(1 - (alpha / 2)));
}
static public double GetBlsRho(double S, double K, double T, double sigma, double r, char type)
{
Accord.Statistics.Distributions.Univariate.NormalDistribution normDist = new Accord.Statistics.Distributions.Univariate.NormalDistribution();
return(type == 'c'?K * T * Math.Exp(-r * T) * normDist.DistributionFunction(D2(S, K, T, sigma, r)):-K *T *Math.Exp(-r *T) * normDist.DistributionFunction(-D2(S, K, T, sigma, r)));
}
static public double GetBlsVega(double S, double K, double T, double sigma, double r)
{
Accord.Statistics.Distributions.Univariate.NormalDistribution normDist = new Accord.Statistics.Distributions.Univariate.NormalDistribution();
return S * Math.Sqrt(T) * normDist.ProbabilityDensityFunction(D1(S, K, T, sigma, r));
}
static public double GetBlsTheta(double S, double K, double T, double sigma, double r, char type)
{
Accord.Statistics.Distributions.Univariate.NormalDistribution normDist = new Accord.Statistics.Distributions.Univariate.NormalDistribution();
return(-normDist.ProbabilityDensityFunction(D1(S, K, T, sigma, r)) * S * sigma / (2 * Math.Sqrt(T)) - r * K * Math.Exp(-r * T) * normDist.DistributionFunction(D2(S, K, T, sigma, r)) * (type == 'c'?1:-1));
}
public static string getReggressionMethod(List <string> typeScale, string typeMethod, List <double> vs)
{
Accord.Statistics.Distributions.Univariate.NormalDistribution normDist =
Accord.Statistics.Distributions.Univariate.NormalDistribution.Standard;
Accord.Statistics.Testing.KolmogorovSmirnovTest kstX =
new Accord.Statistics.Testing.KolmogorovSmirnovTest(vs.ToArray(), normDist);
AnalysisClass.getDistributionScale(kstX.Significant);
switch (typeMethod)
{
case "Простая линейная регрессия":
if ((typeScale[0].Equals("ранговая") || typeScale[0].Equals("количественная")) && typeScale.Count == 2)
{
return("Метод применим к данным");
}
else
{
return("Метод не применим к данным");
}
case "Множественная линейная регрессия":
if ((typeScale[0].Equals("ранговая") || typeScale[0].Equals("количественная")) && typeScale.Count > 2)
{
return("Метод применим к данным");
}
else
{
return("Метод не применим к данным");
}
case "Нелинейная регрессия":
if ((typeScale[0].Equals("ранговая") || typeScale[0].Equals("количественная")) && typeScale.Count >= 2 && AnalysisClass.getDistributionScale(kstX.Significant) == "нормальное")
{
return("Метод применим к данным");
}
else
{
return("Метод не применим к данным");
}
case "Бинарная логистическая регрессия":
if ((typeScale[0].Equals("дихотомическая")) && typeScale.Count > 2)
{
return("Метод применим к данным");
}
else
{
return("Метод не применим к данным");
}
case "Мультиномиальная логистическая регрессия":
if (typeScale[0].Equals("ранговая") && typeScale.Count > 2 && AnalysisClass.getDistributionScale(kstX.Significant) == "нормальное")
{
return("Метод применим к данным");
}
else
{
return("Метод не применим к данным");
}
case "Порядковая регрессия":
if (typeScale[0].Equals("ранговая") && typeScale.Count > 2 && AnalysisClass.getDistributionScale(kstX.Significant) == "нормальное")
{
return("Метод применим к данным");
}
else
{
return("Метод не применим к данным");
}
case "Пробит-анализ":
if ((typeScale[0].Equals("дихотомическая")) && typeScale.Count > 2)
{
return("Метод применим к данным");
}
else
{
return("Метод не применим к данным");
}
default:
return("Нет");
}
}
//static private decimal GetBlsCallPrice(double S, double K, double T, double sigma, double r)
// {
// Accord.Statistics.Distributions.Univariate.NormalDistribution normDist = new Accord.Statistics.Distributions.Univariate.NormalDistribution();
// return decimal.Ceiling((decimal)(S * normDist.DistributionFunction(D1(S, K, T, sigma, r)) - K * Math.Exp(-r * T) * normDist.DistributionFunction(D2(S, K, T, sigma, r)))*100m)/100m;
// //return S * normDist.DistributionFunction(D1(S, K, T, sigma, r)) - K * Math.Exp(-r * T) * normDist.DistributionFunction(D2(S, K, T, sigma, r));
// }
// static private decimal GetBlsPutPrice(double S, double K, double T, double sigma, double r)
// {
// Accord.Statistics.Distributions.Univariate.NormalDistribution normDist = new Accord.Statistics.Distributions.Univariate.NormalDistribution();
// return decimal.Ceiling((decimal)(-S * normDist.DistributionFunction(-D1(S, K, T, sigma, r)) + K * Math.Exp(-r * T) * normDist.DistributionFunction(-D2(S, K, T, sigma, r)))*100m)/100m;
// //return -S * normDist.DistributionFunction(-D1(S, K, T, sigma, r)) + K * Math.Exp(-r * T) * normDist.DistributionFunction(-D2(S, K, T, sigma, r));
// }
// static public decimal GetBlsPrice(double S, double K, double T, double sigma, double r, char type)
// {
// return type == 'c' ? GetBlsCallPrice(S, K, T, sigma, r) : GetBlsPutPrice(S, K, T, sigma, r);
// }
static public double GetBlsDelta(double S, double K, double T, double sigma, double r, char type)
{
Accord.Statistics.Distributions.Univariate.NormalDistribution normDist = new Accord.Statistics.Distributions.Univariate.NormalDistribution();
return(type == 'c'? (normDist.DistributionFunction(D1(S, K, T, sigma, r))) : (normDist.DistributionFunction(D1(S, K, T, sigma, r)) - 1));
}
static public double GetBlsPutPrice(double S, double K, double T, double sigma, double r)
{
Accord.Statistics.Distributions.Univariate.NormalDistribution normDist = new Accord.Statistics.Distributions.Univariate.NormalDistribution();
return -S * normDist.DistributionFunction(-D1(S, K, T, sigma, r)) + K * Math.Exp(-r * T) * normDist.DistributionFunction(-D2(S, K, T, sigma, r));
}
static public double GetBlsRho(double S, double K, double T, double sigma, double r, char type)
{
Accord.Statistics.Distributions.Univariate.NormalDistribution normDist = new Accord.Statistics.Distributions.Univariate.NormalDistribution();
return type=='c'?K * T * Math.Exp(-r * T) * normDist.DistributionFunction(D2(S, K, T, sigma, r)):-K*T*Math.Exp(-r* T)* normDist.DistributionFunction(-D2(S, K, T, sigma, r));
}
private void button2_Click(object sender, EventArgs e)
{
SaveRegression saveRegression = new SaveRegression();
if (!string.IsNullOrEmpty(richTextBox1.Text))
{
string NumberOfVariable = "", NumberOfUqVariable = "", TypeOfScale = "", DistType = "", AGV = "", Slope = "", Mechanics = "", StandardDeviation = "";
List <double> listtemp = new List <double>();
int temp = 0;
Accord.Statistics.Distributions.Univariate.NormalDistribution normDist =
Accord.Statistics.Distributions.Univariate.NormalDistribution.Standard;
double[,] matr = new double[dataGridView1.RowCount, dataGridView1.ColumnCount];
for (int j = 0; j < dataGridView1.ColumnCount; j++)
{
for (int i = 0; i < dataGridView1.RowCount; i++)
{
matr[i, j] = Convert.ToDouble(dataGridView1.Rows[i].Cells[j].Value);
if (!listtemp.Contains(matr[i, j]))
{
temp++;
}
if (dataGridView1.Rows[i].Cells[j].Value != null)
{
listtemp.Add(matr[i, j]);
}
}
TypeOfScale += AnalysisClass.getScale(temp) + "!";
NumberOfUqVariable += temp.ToString() + "!";
NumberOfVariable += listtemp.Count().ToString() + "!";
if (AnalysisClass.getScale(temp) != "дихотомическая")
{
Accord.Statistics.Testing.KolmogorovSmirnovTest kstX =
new Accord.Statistics.Testing.KolmogorovSmirnovTest(listtemp.ToArray(), normDist);
DistType += AnalysisClass.getDistributionScale(kstX.Significant) + "!";
AGV += Math.Round(MathNet.Numerics.Statistics.Statistics.Mean(listtemp), 3).ToString() + "!";
StandardDeviation += Math.Round(MathNet.Numerics.Statistics.Statistics.PopulationStandardDeviation(listtemp), 3).ToString() + "!";
Mechanics += Math.Round(MathNet.Numerics.Statistics.Statistics.PopulationSkewness(listtemp), 3).ToString() + "!";
Slope += Math.Round(MathNet.Numerics.Statistics.Statistics.PopulationKurtosis(listtemp), 3).ToString() + "!";
}
else
{
DistType += "Нет!";
AGV += "Нет!";
StandardDeviation += "Нет!";
Mechanics += "Нет!";
Slope += "Нет!";
}
temp = 0;
listtemp.Clear();
}
DateRegression.TypeMethod = richTextBox1.Text;
DateRegression.NumberOfScale = dataGridView1.ColumnCount;
DateRegression.NumberOfVariable = NumberOfVariable;
DateRegression.NumberOfUqVariable = NumberOfUqVariable;
DateRegression.TypeOfScale = TypeOfScale;
DateRegression.DistType = DistType;
DateRegression.StandardDeviation = StandardDeviation;
DateRegression.Mechanics = Mechanics;
DateRegression.AGV = AGV;
DateRegression.Slope = Slope;
saveRegression.Show();
}
else
{
MessageBox.Show("Сначало нужно нажать кнопку 'Предложить метод исследования'");
}
}
static public double GetBlsPutPrice(double S, double K, double T, double sigma, double r)
{
Accord.Statistics.Distributions.Univariate.NormalDistribution normDist = new Accord.Statistics.Distributions.Univariate.NormalDistribution();
return(-S *normDist.DistributionFunction(-D1(S, K, T, sigma, r)) + K * Math.Exp(-r * T) * normDist.DistributionFunction(-D2(S, K, T, sigma, r)));
}
static public double GetBlsVega(double S, double K, double T, double sigma, double r)
{
Accord.Statistics.Distributions.Univariate.NormalDistribution normDist = new Accord.Statistics.Distributions.Univariate.NormalDistribution();
return(S * Math.Sqrt(T) * normDist.ProbabilityDensityFunction(D1(S, K, T, sigma, r)));
}
static public double GetBlsTheta(double S, double K, double T, double sigma, double r, char type)
{
Accord.Statistics.Distributions.Univariate.NormalDistribution normDist = new Accord.Statistics.Distributions.Univariate.NormalDistribution();
return -normDist.ProbabilityDensityFunction(D1(S, K, T, sigma, r))*S*sigma / (2 * Math.Sqrt(T))-r*K*Math.Exp(-r*T)*normDist.DistributionFunction(D2(S, K, T, sigma, r))*(type=='c'?1:-1);
}
