Digging insights out of multivariate data is intrinsically hard. Data Modeler’s evolutionary algorithms address this by rewarding simplicity as well as accuracy to identify the most impactful variables and variable combinations as well as to develop descriptive models. The stochastic nature of the model search also means that alternate possibilities are explored and refined. The diversity of developed models form the foundation for models with trust metrics to detect extrapolation or system changes as well as to guide future data collection for maximum information value. In this talk we will demonstrate these concepts using the Data Modeler GUI and illustrate how letting the data speak for itself can be used for real world applications such as chemical plant troubleshooting and optimization.
For the newest resources, visit Wolfram Repositories and Archives »
