R-squared vs number of basis functions plot

Note

This command is available with the Predictive Analytics Module. Click here for more information about how to activate the module.

The R-squared vs number of basis functions plot displays R² values on the y-axis and the number of basis functions on the x-axis. The R² value indicates whether the model is a good fit. When the analysis uses a validation method, the plot includes a line for the validation results. Use the validation results to assess the performance of the model to predict new observations. Compare the training results and the validation results to see whether there are overfitting problems with the model for the training data set.

The largest x-value for a point shows that this analysis evaluates 20 basis functions. The optimal number of basis functions is 13. The R² value for the test data when the number of basis functions is 13 is approximately 87.61%.

When the absolute deviation loss function determines the optimal model, then the results include the MAD vs number of basis functions plot instead.

Interpretation

Higher values of R² indicate a better model. The reference line indicates the optimal R² value for the validation results and the number of basis functions in the model. If the validation curve indicates an insufficient model, consider whether to retry the analysis with alternative settings, such as a search of more basis functions.