Check Accuracy

The R² Accuracy check is applicable only for static models.

In the case of dynamic models to evaluate its prediction accuracy you may use the RMSE metric available in the Time Simulation tab.

Select the Accuracy check radio button.

Accuracy check requires a test dataset for prediction. If the test dataset is not provided or if it incompatible with the model feature names, then an error dialog box is shown.

The test dataset name is populated in the file name info entries on the bottom of the panel.

The training dataset is automatically loaded as test dataset in the post-processor just after the training. In this case, the training data points are shown with blue color and test data points are shown with green color.

The y-axis shows the predictions of the selected output produced by the romAI model, given the inputs defined in the dataset file.

The x-axis shows the targets for the same selected output contained in the same dataset file.

The closer that the predictions are to the targets, accordingly, more blue points will be aligned to the y = x curve, drawn in red.

The R² coefficient gets computed automatically.