Generalization of the mechanisms of cross-correlation analysis in the case of a multivariate time series

The article describes a generalization of the mechanisms of cross-correlation analysis in the case of a multivariate time series and how this allows the optimal lags to be identified for each of the independent variables (IV) using a number of algorithms. The use of generalized mechanisms will allow variables to be analysed and predicted based on the retrospective analysis of multidimensional data. In the available literature, cross-correlation has been defined only for pairs of time series. However, the study of dependent variable (DV) dependencies on multidimensional independent variables that takes into account the vector of specially selected time lags will significantly improve the quality of models based on multiple regression. The idea of multiple cross-correlation lies in the sequential forward shift of each IV row with respect to DV (it transpires that DV is delayed relative to IV) until we obtain a minimum error or the best test of multiple regression. After the completion of all stages of multiple cross-correlation, the synthesis of the model is not a difficult process.