You can use and analyze the best model resulting from automatic model optimization in a multitude of different ways:
For example, prediction: For new parameter settings (e.g., process parameter settings, product configurations) the model output (quality, stability, cost, …) can be predicted. Moreover, you can analyze models for parameter sensitivities, understanding the impact of variation of one or two parameters on the model output. Using ternary diagrams, you can also investigate the impact of three parameters which are summing up to a constant value. This is a frequently used application in formulation design.
Through model analysis, you can also inspect the model for various model types. For example, linear models and fuzzy models (inspection of rules) allow for this type of model analysis, and often this analysis facilitates an enhanced understanding of complex relationships in the underlying real-world system.
Additional helpful analysis features include the determination of parameter importance (e.g., by means of so-called Sobol-indices) and utilization of parameter dendrograms for dimensionality reduction. Both methods can be used easily by means of graphical visualizations.
+49 231 97 00 342
Thomas Bäck will introduce research results at the GECCO (Genetic and Evolutionary...
Case studies give you an inspiration for possible optimization solutions in your own company.
divis intelligent solutions GmbH
+49 231 97 00 341