procedure ARBurgFit(const Data: TVec; const Phi: TVec; out Sigma2: Double; const StdErrs: TVec);
Burg AR estimation.
| # | Name | Description |
|---|---|---|
| 1 | Data | Zero-mean time series. If this is not the case, subtract the mean from data. |
| 2 | Phi | Returns estimates for Phi coefficients. AR(p) order is determined by Phi length. |
| 3 | Sigma2 | Returns Burg estimated variance for AR process. |
| 4 | StdErrs | Returns estimated phi coefficients standard errors. |
Result: stored in self (calling object)
Remarks:
Performs Burg estimation for pure (AR) model.
Examples
Uses MtxExpr, StatTimeSerAnalysis, Math387;
procedure Example;
var ts,phi,stdErr: Vector;
s2: double;
begin
ts.LoadFromFile('timeser.vec');
phi.Length := 3; // for AR(3) process
ARBurgFit(ts,phi,s2,stdErr);
end;
See Also: StatTimeSerAnalysis.ARYuleWalkerFit