We know that red noise is the same as a first-order autoregressive (AR(1)) stationary Gaussian process with a positive correlation at unit lag. Does there exist some formal statistical test for red noise process?
One way I can think of is to firstly fit an AR(1) model to the process, then perform white noise test on the model residuals to see whether the residuals follow a white noise process. If yes, the original process can be viewed as a red noise process...
Welcome any ideas and comments.
