To expand the assessment to include information from all time series, we require the use of multiEWS . The single difference in the function’s parameterisation is that the input data frame must contain more than two columns (one time sequence column and two or more time series). By default, all indicators are returned.
To compute multivariate EWSs using the rolling method, the function would be written as thus, specifying the method and winsize as a percentage of the time series’ length: