New Options to use Multiple dimensions for Calibration – Version 5.3.5

The new version allows your calibration to be by using one dimension for fixed variability of parameters
and the other dimensions for random selection of parameter values.

The main purpose is to use the first dimension for variability by coordinated or because of any experimental treatment that could be understood by various a fixed predetermined variability. The Unknown dimension from 2-n could be assigned to meet your interest for estimating which parameters that can be represented by common values to best describe the variability for the first dimension.

The performance for the Variability within the first dimension will be described by those Single Variable Criteria that can also be assigned after you have completed the multirun.
Note that previous Single Variable Validation was described by only ME, RMSE and LogLi since they described the difference for a single variable and only one single run.

The new procedure will use Single Variable Validation defined in higher dimension of 1 and the performance to describe the variability for all runs within the first dimesion by using
all those indicator normally are used for time serie validation. This mean that R2, Intercept, Slope, ME, RMSE, LogLi and NSE -R2. Note that you can define criteria for selection
of valid runs for all peformance indicators. The peformance indicators for higher dimension validation variables will be repeated for the number of runs within the first dimension.

More complete description with an example for how to use the new procedure for Calibration will follow soon. If you by some reasons get problems with some previous multi-run calibration using only
one dimension when using the new version – Please let me know.

Leave a Reply

Your email address will not be published.