Symbolic Regression

Symbolic regression is a type of regression analysis on a symbolic level. Transferred to an application in the context of ASCMO-MOCA, this means to automatically find an equation-based or hybrid (mixing equation- and data-based approaches) model with the following properties:

Model corresponds to a dataset well in a statistical sense.
Model is as compact as desired.
Model is human-interpretable.

The symbolic regression plugin of ASCMO-MOCA provides a solution to this task by carrying out optimizations on the structural level of equations and local optimizations to fit identified models to data. In terms of embedded software function engineering these two steps correspond to function engineering and calibration, respectively.

ASCMO-MOCA supports engineers carrying out this steps more efficiently and effectively using artificial-intelligence. From an alternative perspective, the method can also be viewed as an automated way of system identification.

The Symbolic Regression Feature is located in the Function Step.