Abstract:
In this paper, we consider processes of the MARMA type that can be derived from the classical ARMA processes by replacing summation by the maximum operation. It is assumed that the innovations and the values of the process have the standard Fréchet distribution. For simple MARMA processes of first order, certain numerical characteristics are calculated. Sign tests and rank statistical methods for parameter estimation are developed. The characterization relations that can be used for the identification of models are justified.
Keywords:
MARMA and ARMA processes, moving maximum, sign test, rank test, estimation of parameters, random variable, Fréchet distribution, Pearson correlation coefficient.
This publication is cited in the following 8 articles:
A. V. Lebedev, “Multivariate Extremes of Random Scores of Particles in Branching Processes with Max-Linear Inheritance”, Math. Notes, 105:3 (2019), 376–384
A. V. Lebedev, “Statisticheskii analiz maksimum-lineinykh sluchainykh protsessov”, Sistemy i sredstva inform., 27:2 (2017), 16–28
Ferreira M., “the Lawrence-Lewis Pareto Process: An Extremal Approach”, Electron. J. Appl. Stat. Anal., 9:1 (2016), 68–82
Ferreira M., Ferreira H., “Extremes of Multivariate Armax Processes”, Test, 22:4 (2013), 606–627
Ferreira M., “On the extremal behavior of a Pareto process: an alternative for ARMAX modeling”, Kybernetika (Prague), 48:1 (2012), 31–49
M. Ferreira, “On Tail Dependence: A Characterization for First-Order Max-Autoregressive Processes”, Math. Notes, 90:6 (2011), 882–893
Ferreira M., “Estimation of the parameter of a pARMAX model”, REVSTAT, 8:2 (2010), 139–149
A. V. Lebedev, “Nonlinear Prediction in Max-Autoregressive Processes”, Math. Notes, 85:4 (2009), 602–606