Eureqa Desktop User Guide

Viewing Model Results in Eureqa Desktop

This is where you view and compare the best solutions Eureqa has found thus far.



The best solutions are determined by two factors: their complexity ("Size") and their accuracy ("Fit") on the validation data. Those listed here have the highest accuracy for various complexities/sizes. By default, Eureqa ranks solutions according to a ratio of complexity and accuracy; solutions that are accurate but not too complex are shown at the top. You can re-sort by complexity alone or by accuracy alone by clicking on "Size" or "Fit". Select any formula (by clicking) or group of formulae (by Ctrl-clicking or clicking and dragging) to see the selected formula or formulae plotted and analyzed.

return to top ⇑

Solution Plot

Selected solutions are plotted against the data. Both training and validation data are shown. The training data is a subset of your data that is used by Eureqa to search for solutions. The validation data is a second subset that is used only to measure accuracy.

In addition to the solution plot, several other plots are available:

  • Observed vs. Predicted Plot - Plots actual observed values which appear in the training and validation data against the values predicted by the selected model.
  • Residual Error Plot - Plots the size of the error for each point in the training and validation data.
  • ROC, Sensitivity/Specificity Plot - Plots the ROC Curve for the data. Use for classification models which predict target variables that always have a value of 0 or 1.
return to top ⇑

Solution Details

View the performance metrics of selected solutions. When multiple solutions are selected, their performance can be compared side by side.

return to top ⇑

Pareto Chart

View the solutions as they rank by accuracy and complexity. Complex but accurate solutions will lie at the lower right, while simple but inaccurate solutions will lie at the top left. The most useful solutions are usually somewhere in the middle, striking the right balance between complexity and accuracy.

return to top ⇑