Time Series Analysis: Forecasting and Control
Time Series Analysis: Forecasting and Control
| Hi-index | 0.00 |
Consideration was given to the problem of cluster analysis of observations which are realizations of the autoregression time series that are classified according to the unknown values of the autoregression parameters (coefficients). To solve the problem of classification, constructed were a procedure of cluster analysis in the space of realizations and a cluster procedure in the space of maximum likelihood estimates of the autoregression parameters. Efficiency of the proposed cluster procedures was studied both analytically and experimentally.