ISmLinearRegress.GLSMatrix

Syntax

GLSMatrix: Array;

Description

The GLSMatrix property determines a covariance matrix for generalized least squares method.

Comments

A covariance matrix should be symmetric and positive definite.

Example

Add a link to the Stat system assembly.

Sub UserProc;
Var
    OLS: ISmLinearRegress;
    can, fra, ger: Array[10] Of Double;
    omega: Array[10, 10] Of Double;
    res: Integer;
Begin
    OLS := New SmLinearRegress.Create;
    // Set values for variables
    can[0] := 6209; fra[0] := 4110; ger[0] := 3415;
    can[1] := 6385; fra[1] := 4280; ger[1] := 3673;
    can[2] := Double.Nan; fra[2] := 4459; ger[2] := 4013;
    can[3] := 6837; fra[3] := 4545; ger[3] := 4278;
    can[4] := 6495; fra[4] := 4664; ger[4] := 4577;
    can[5] := 6907; fra[5] := 4861; ger[5] := 5135;
    can[6] := 7349; fra[6] := 5195; ger[6] := 5388;
    can[7] := 7213; fra[7] := 5389; ger[7] := 5610;
    can[8] := 7061; fra[8] := 5463; ger[8] := 5787;
    can[9] := 7180; fra[9] := 5610; ger[9] := 6181;
    omega[1, 1] := 0;
    omega[2, 2] := 0;
    omega[5, 5] := 0;
    // Sample period parameters
    OLS.ModelPeriod.FirstPoint := 1;
    OLS.ModelPeriod.LastPoint := 10;
    // Choose explained variable
    OLS.Explained.Value := can;
    // Choose regressors
    OLS.Explanatories.Clear;
    OLS.Explanatories.Add.Value := fra;
    OLS.Explanatories.Add.Value := ger;
    // Parameters of model coefficients
    OLS.ModelCoefficients.Intercept.Mode := InterceptMode.AutoEstimate;
    // Method of missing data treatment
    OLS.MissingData.Method := MissingDataMethod.LinTrend;
    // Value of weights use
    OLS.UseWeights := True;
    // Setting covariance matrix for generalized least squares method
    OLS.GLSMatrix := omega;
    // Type of least squares method
    OLS.LSType := LRLSType.GLS;
    res := OLS.Execute;
    Debug.WriteLine(OLS.Errors);
    Debug.WriteLine("Summary statistics");
    Debug.Indent;
Debug.WriteLine("Fisherstatistic:"+OLS.SummaryStatistics.Fstat.ToString);     Debug.WriteLine("Probability for Fisher statistics: " + OLS.SummaryStatistics.ProbFstat.ToString);
    Debug.Unindent;
End Sub UserProc;

As a result of the example execution, the following settings are defined:

Sample period.
Method of missing data treatment.
Parameters of model coefficients.
Covariance matrix for generalized least squares method.
Type of least squares method.

The console window shows summary statistics.