Add: INonLinearRegressionExplanatory;
Add: Prognoz.Platform.Interop.Stat.INonLinearRegressionExplanatory;
The Add method adds an explanatory series to the collection.
Add a link to the Stat system assembly.
Sub UserProc;
Var
NLr: ISmNonLinearLeastSquare;
ar: Array[0..6] Of Double;
ar2: Array[0..8] Of Double;
ar3: Array[0..8] Of Double;
InitEst: Array[0..1] Of Double;
Per: IStatPeriod;
res, i: Integer;
MC: ISlConstCoefficients;
Regs: INonLinearRegressionExplanatories;
Begin
NLr := New SmNonLinearLeastSquare.Create;
// Set values for variables
ar[0] := 100; ar2[0] := 120; ar3[0] := 122;
ar[1] := 111; ar2[1] := 125; ar3[1] := 127;
ar[2] := 123; ar2[2] := 124; ar3[2] := 130;
ar[3] := 113; ar2[3] := 130; ar3[3] := 135;
ar[4] := 119; ar2[4] := 133; ar3[4] := 140;
ar[5] := 121; ar2[5] := 129; ar3[5] := 149;
ar2[6] := 139; ar3[6] := 150;
ar2[7] := 140; ar3[7] := 155;
// Set explained variable
NLr.Explained.Value := ar;
// Determine parameters of regressor collection for non-linear OLS
Regs := NLr.Explanatories;
// Add explanatory series to the collection
Regs.Add.Serie.Value := ar2;
Regs.Add.Serie.Value := ar3;
// Determine calculated function
NLr.FunctionString := "A1 + A2 * " + Regs.Item(0).VariableName + " * " + Regs.Item(1).VariableName;
// Determine coefficient order
NLr.CoefficientsOrder := "A1;A2";
// Define parameters of sample period
Per := NLr.ModelPeriod;
Per.FirstPoint := 1;
Per.LastPoint := 5;
// Determine forecasting series parameters
NLr.Forecast.LastPoint := 9;
// Determine maximum number of iterations
NLr.MaxIteration := 100;
// Determine whether analytical derivatives are used on solution search
NLr.UseDerivatives := False;
// Determine accuracy
NLr.Tolerance := 0.003;
// Determine initial approximations
InitEst[0] := 0.5;
InitEst[1] := 1.0;
Nlr.InitApproximation := InitEst;
// Run calculation and show results
res := NLr.Execute;
If res <> 0 Then
Debug.WriteLine(NLr.Errors);
Else
Debug.WriteLine("== Coefficients ==");
Debug.WriteLine("A1");
MC := NLr.ModelCoefficients("A1");
Debug.WriteLine("value: " + MC.Estimate.ToString);
Debug.WriteLine("probability: " + MC.Probability.ToString);
Debug.WriteLine("A2");
MC := NLr.ModelCoefficients("A2");
Debug.WriteLine("value: " + MC.Estimate.ToString);
Debug.WriteLine("probability: " + MC.Probability.ToString);
Debug.WriteLine("== Modeling series ==");
For i := 0 To NLr.Fitted.Length - 1 Do
Debug.WriteLine(NLr.Fitted[i]);
End For;
End If;
End Sub UserProc;
After executing the example the console window displays calculation results: coefficients, their values and probabilities, and a modeling series.
The requirements and result of the Fore.NET example execution match with those in the Fore example.
Imports Prognoz.Platform.Interop.Stat;
…]
Public Shared Sub Main(Params: StartParams);
Var
NLr: ISmNonLinearLeastSquare;
ar: Array[0..6] Of Double;
ar2: Array[0..8] Of Double;
ar3: Array[0..8] Of Double;
InitEst: Array[0..1] Of Double;
Per: IStatPeriod;
res, i: Integer;
MC: ISlConstCoefficients;
Regs: INonLinearRegressionExplanatories;
Fitted: System.Array;
Begin
NLr := New SmNonLinearLeastSquare.Create();
// Set values for variables
ar[0] := 100; ar2[0] := 120; ar3[0] := 122;
ar[1] := 111; ar2[1] := 125; ar3[1] := 127;
ar[2] := 123; ar2[2] := 124; ar3[2] := 130;
ar[3] := 113; ar2[3] := 130; ar3[3] := 135;
ar[4] := 119; ar2[4] := 133; ar3[4] := 140;
ar[5] := 121; ar2[5] := 129; ar3[5] := 149;
ar2[6] := 139; ar3[6] := 150;
ar2[7] := 140; ar3[7] := 155;
// Set explained variable
NLr.Explained.Value := ar;
// Determine parameters of regressor collection for non-linear OLS
Regs := NLr.Explanatories;
// Add explanatory series to the collection
Regs.Add().Serie.Value := ar2;
Regs.Add().Serie.Value := ar3;
// Determine calculated function
NLr.FunctionString := "A1 + A2 * " + Regs.Item[0].VariableName + " * " + Regs.Item[1].VariableName;
// Determine coefficient order
NLr.CoefficientsOrder := "A1;A2";
// Define parameters of sample period
Per := NLr.ModelPeriod;
Per.FirstPoint := 1;
Per.LastPoint := 5;
// Determine forecasting series parameters
NLr.Forecast.LastPoint := 9;
// Determine maximum number of iterations
NLr.MaxIteration := 100;
// Determine whether analytical derivatives are used on solution search
NLr.UseDerivatives := False;
// Determine accuracy
NLr.Tolerance := 0.003;
// Determine initial approximations
InitEst[0] := 0.5;
InitEst[1] := 1.0;
Nlr.InitApproximation := InitEst;
// Run calculation and show results
res := NLr.Execute();
If res <> 0 Then
System.Diagnostics.Debug.WriteLine(NLr.Errors);
Else
System.Diagnostics.Debug.WriteLine("== Coefficients ==");
System.Diagnostics.Debug.WriteLine("A1");
MC := NLr.ModelCoefficients["A1"];
System.Diagnostics.Debug.WriteLine("value: " + MC.Estimate.ToString());
System.Diagnostics.Debug.WriteLine("probability: " + MC.Probability.ToString());
System.Diagnostics.Debug.WriteLine("A2");
MC := NLr.ModelCoefficients["A2"];
System.Diagnostics.Debug.WriteLine("value: " + MC.Estimate.ToString());
System.Diagnostics.Debug.WriteLine("probability: " + MC.Probability.ToString());
System.Diagnostics.Debug.WriteLine("== Modeling series ==");
Fitted := NLr.Fitted;
For i := 0 To NLr.Fitted.Length - 1 Do
System.Diagnostics.Debug.WriteLine(Fitted[i]);
End For;
End If;
End Sub;
See also: