ISmPrincipalComponentAnalysis.CalcCovariance

Syntax

Description

The CalcCovariance property determines the method used to calculate the standard deviation.

Comments

The value of this property is used only if type of source data analysis is selected as Correlation matrix of source data (ISlAnalysis.Type).

By default, the standard deviation is calculated based on a universal set (GeneralCovariance).

Example

Sub Main;

Var

pc: SmPrincipalComponentAnalysis;

Obj: ISlSeries;

d0, d1, d2: Double;

i, res: Integer;

ar1,ar2,ar3: Array Of Double;

Begin

pc := New SmPrincipalComponentAnalysis.Create;

pc.MissingData.Method := MissingDataMethod.AnyValue;

ar1 := New Double[10];

ar2 := New Double[10];

ar3 := New Double[10];

// values ar1, ar2, ar3

ar1[0] := 7.0000; ar2[0] := 4.0000; ar3[0] := 3.0000;

ar1[1] := 4.0000; ar2[1] := 1.0000; ar3[1] := 8.0000;

ar1[2] := 6.0000; ar2[2] := 3.0000; ar3[2] := 5.0000;

ar1[3] := 8.0000; ar2[3] := 6.0000; ar3[3] := 1.0000;

ar1[4] := 8.0000; ar2[4] := 5.0000; ar3[4] := 7.0000;

ar1[5] := 7.0000; ar2[5] := 2.0000; ar3[5] := 9.0000;

ar1[6] := 5.0000; ar2[6] := 3.0000; ar3[6] := 3.0000;

ar1[7] := 9.0000; ar2[7] := 5.0000; ar3[7] := 8.0000;

ar1[8] := 7.0000; ar2[8] := 4.0000; ar3[8] := 5.0000;

ar1[9] := 8.0000; ar2[9] := 2.0000; ar3[9] := 2.0000;

Obj := pc.Objects;

Obj.Add.Value := ar1;

Obj.Add.Value := ar2;

Obj.Add.Value := ar3;

pc.Analysis.Type := AnalysisType.Correlation;

pc.CalcCovariance := CalcCovarianceType.SampleCovariance;

pc.ScoreType := PCAScoreType.Stand;

res := pc.Execute;

If res <> 0 Then

Debug.WriteLine(pc.Errors);

Else

Debug.WriteLine("Eigenvectors: ");

For i := 1 To 3 Do

d0 := pc.P[0, i - 1];

d1 := pc.P[1, i - 1];

d2 := pc.P[2, i - 1];

Debug.AssertMsg(False, "Principal component №" + i.ToString + ": " + d0.ToString + " " + d1.ToString + " " + d2.ToString);

End For;

Debug.WriteLine("================================");

Debug.WriteLine("Principal component values: ");

Debug.AssertMsg(False, "Principal component №1, №2, №3");

For i := 1 To pc.v.GetUpperBound(1) + 1 Do

d0 := pc.V[i - 1, 0];

d1 := pc.V[i - 1, 1];

d2 := pc.V[i - 1, 2];

Debug.AssertMsg(False, "Observation №" + i.ToString + ": " + d0.ToString + " " + d1.ToString + " " + d2.ToString);

End For;

End If;

End Sub Main;

After executing the example the console window displays eigenvectors and values of principal components:

Module execution started

Eigenvectors:

Principal component №1: 0.64200457634989416 0.68636164136057776 -0.34166916924798257

Principal component №2: 0.38467229168844341 0.097130330134305665 0.91792860668744924

Principal component №3: -0.66321742434359598 0.72074502858973277 0.20166618906061454

================================

Principal component values:

Principal component №1, №2, №3

Observation 1: 0.12903035852981545 -0.21839226959071706 0.02025485072666143

Observation 2: -0.66669301100579936 0.0217442534129067 0.19999666028421142

Observation 3: -0.14638086480259591 -0.10060632857139759 0.094644662986802838

Observation 4: 0.51324074516097307 -0.31491122407391198 0.22138084765686122

Observation 5: 0.22137977909251411 0.34314958971596865 0.20643080180376572

Observation 6: -0.27162977083739037 0.41840155766303921 -0.27021100431677547

Observation 7: -0.19095172298978533 -0.41443860123329074 0.27083977554578842

Observation 8: 0.29646093774132237 0.54376056232352743 -0.01319193797809412

Observation 9: 0.068009757606577848 0.0080503305179512527 0.10711010517240993

Observation 10: 0.047533791504367734 -0.28675787016407545 -0.83725476188162995

Module execution finished