otkerneldesign
TestSetWeighting¶
- class otkerneldesign.TestSetWeighting(train, test, distribution_sample, kernel=None)¶
Weighting of a test-set for optimal machine learning performance metric (e.g, Integrated Squared Error, predictivity coefficient) estimation.
- Parameters:
- train
openturns.Sample Training set used for fitting a machine learning model.
- test
openturns.Sample Training set used for validating a machine learning model.
- distribution_sample2-d list of float
Large sample that empirically represents a distribution.
- kernel
openturns.CovarianceModel Covariance kernel used to define potentials. By default a product of Matern kernels with smoothness 5/2.
- train
Examples
TODO
Methods
- compute_weights(residuals=None)¶
Compute optimal-weights for better performance estimation of a machine learning model.
- Parameters:
- residuals: list
List of residuals in the case of a non-interpolating machine learning model. By default set to None for an interpolating model (e.g., Gaussian process regression).