pimkl.models package

Submodules

pimkl.models.pimkl module

Pathway Induced Multiple Kernel Learning.

class pimkl.models.pimkl.PIMKL(inducers, induction='induce_linear_kernel', mkl='UMKLKNN', estimator='EasyMKL', induction_parameters={}, mkl_parameters={'epsilon': 0.0001, 'k': 5, 'kernel_normalization': True, 'maxiter_qp': 100000, 'precompute': True}, estimator_parameters={'epsilon': 1e-05, 'kernel_normalization': False, 'lam': 0.2, 'precompute': True, 'regularization_factor': False})

Bases: BaseEstimator, ClassifierMixin

Pathway Induced Multiple Kernel Learning with choice of MKL and estimator algorithm. Estimator is only trained when MKL is not an estimator itself.

fit(X, y=None)

Fit the model. Estimator is only trained when MKL is not an estimator.

get_params(deep=True)

Get model parameters.

predict(X)

Predict using trained model.

It returns the optimal kernel using learned weights or, in case labels were fitted in training, the predicted labels.

predict_proba(X)

Predict probabilities using trained model.

set_estimator_params(**parameters)

Set model parameters.

set_mkl_params(**parameters)

Set model parameters.

set_params(**parameters)

Set model parameters.

Module contents