I am quite new in this field, but I hope someone can help me regarding my current task. In terms of signal processing, I would like to model one signal which has a random (stochastic) characteristic, which can't be expressed in a closed mathematical form. Due to that problem, I read about time series analysis, but I didn't understand how to apply the method to my problem.
Is it correct that I can model any stochastic stationary process as an AR,MA or ARMA model with respect to autocorrelation and partial autocorrelation?
If yes, how can I check the stationarity of a stochastic process?
After differentiation of a non-stationary process, how can I build the model with integration?
What is the optimal method for estimating the parameters of the model?
I know I am asking many questions, but I am sure that someone can help me.
Kind Regards, Sam