ClearVu Analytics Modeling

Optimal model

Generate the best possible model automatically

The modeling approach of ClearVu Analytics (CVA) provides you with the capability to automatically learn models from data. ClearVu Analytics applies a multitude of (nonlinear) modeling algorithms to the data, because it is not possible to know beforehand , which method will perform best. 

For all modeling algorithms, ClearVu Analytics optimizes the corresponding learning parameters. This approach automatically guarantees application of the best algorithm and the ultimate generation of the best model for your data set. As an evaluation criterion for models their prediction accuracy is used. This approach completely avoids the often occurring phenomenon of overfitting of models.

After the automatic modeling process, a simple traffic light scheme guides you through the results and visualizes whether a good model was found and which modeling algorithm won the competition. You can directly proceed to use this model for prediction, sensitivity analysis, or optimization applications.

Model tree showing model quality evaluation results
Model tree showing model quality evaluation results
Scatterplot for comparing the predictive quality of different models
Scatterplot for comparing the predictive quality of different models
Example of a fuzzy model learned on a data set
Example of a fuzzy model learned on a data set


While ClearVu Analytics provides you with all modern methods of nonlinear data analysis, generating the optimal model is still very simple for the user. It just requires some computational effort. The results are presented to you in a comprehensible way. Among the available modeling algorithms are:

  • generalized linear models
  • support vector machines
  • fuzzy models
  • decision trees
  • ensembles of decision trees (“random forests”)
  • neural networks (MLP, feed-forward)
  • Gaussian processes
  • partial least squares regression (PLS)
  • principle component analysis

As an expert user, you can also set the learning parameters of the models by yourself – however, we recommend not doing so due to the power of the available automatic modeling technology.

Contact

+49 231 97 00 342

Breaking News

30.03.2016

Industrie 4.0

divis contributes to the development of networked systems

Case studies

Case studies give you an inspiration for possible optimization solutions in your own company.

Read more

Customers login

Please log in, you will be redirected automatically.

Contact details

divis intelligent solutions GmbH
Joseph-von-Fraunhofer-Str. 20
44227 Dortmund
+49 231 97 00 341
contact(at)divis-gmbh.com