1: \begin{abstract}

2: %FIXME refaire

3: 

4: We tackle the issue of the blind prediction of a Gaussian time series. For this, we construct a projection operator build by plugging an empirical covariance estimation into a Schur complement decomposition of the projector. This operator is then used to compute the predictor. Rates of convergence of the estimates are given.

5: 

6: % 

7: % 

8: % In the frame of time series analysis,

9: % we compute the quadratic error in the blind estimation of the projection operator for prediction with infinite past. The estimation is made using only a single finite sample of the time series. It is performed by plugging the empirical covariance in a Schur complement decomposition  of the projector. 

10: %Under assumptions on the regularity of the spectral density, we provide the rate of convergence for the error. 

11: \end{abstract}

12: