On detection of feedback in the time series

Abstract EN

This research has characterized is the detection of feedback between output series and input series.

The importance of detective appearance of feedback when has the adequate model is known, it also has applied to some method for detecting the feedback by using two methods.

The first method is testing of cross-correlation function between prewhitening to the input and output series, where it has founded cross-correlation between prewhitening of the input series ( t ) and residual series (at) .The second method is uses to F test between two series of input and output.

It is appearing that the Dynamical Regression models have no adequacy in feedback, then the Multivariate Autoregressive should be used in the case.