Weakly Consistent Offline Clustering of ARMA Processes
Garik Adamyan · Journal of Contemporary Mathematical Analysis (Armenian Academy of Sciences) · 2023
Abstract In this paper, we consider the problem of weakly consistent offline clustering of ARMA processes. Under the provided assumptions we derive a weakly consistent clustering algorithm of invertible ARMA processes according to their forecast functions. Using BIC penalized quasi-maximum likelihood estimate of the distance function the weak consistency of Algorithm 1 is proven when the target number of clusters is known. The theoretical lower bound of the clustering function is provided.