GetParetoFront: Pareto Front data for a Eureqa model

View source: R/ParetoFront.R

GetParetoFrontR Documentation

Pareto Front data for a Eureqa model


The Eureqa algorithm generates millions and millions of equations. Eureqa takes the best bits from the best initial models and splices them randomly into the next generation. After enough mixing, the models can achieve good accuracy. There are usually many equations at every complexity level, but they aren't exposed. The models that are displayed are the "Pareto-optimal" models. That means that for any given complexity score, it shows the model with the best error metric on the training data out of all the modes. After that, for each remaining model, if there a strictly better model, throw out the strictly-worse model. A Pareto Front are those "Pareto-optimal" models that are generated at various complexity scores.





An S3 object of class dataRobotModel like that returned by the function GetModel, or each element of the list returned by the function ListModels.


data.frame with the following components:

  • projectId character. the id of the project the model belongs to

  • errorMetric character. Eureqa error-metric identifier used to compute error metrics for this search. Note that Eureqa error metrics do NOT correspond 1:1 with DataRobot error metrics – the available metrics are not the same, and even equivalent metrics may be computed slightly differently.

  • hyperparameters list. A list of the various hyperparameters that could be used. By default there are none.

  • targetType character. Indicating what kind of modeling is being done in this project Options are: "Regression", "Binary" (Binary classification), "Multiclass" (Multiclass classification)

  • solutions list. List of Pareto points. Every Pareto point contains a dictionary with keys:

    • eureqaSolutionId character. ID of this solution

    • complexity numeric. Complexity score for this solution. Complexity score is a function of the mathematical operators used in the current solution. The Complexity calculation can be tuned via model hyperparameters.

    • error numeric. Error for the current solution, as computed by Eureqa using the "error_metric" error metric.

    • expression character. String specifying the Eureqa model equation.

    • expression_annotated character. Eureqa model equation string with variable names tagged for easy identification.


## Not run: 
  projectId <- "5b2827556523cd05bd1507a5"
  modelId <- "5b29406c6523cd0665685a8d"
  model <- GetModel(projectId, modelId)

## End(Not run)

datarobot documentation built on Aug. 8, 2023, 1:10 a.m.